An electromagnetic compatibility management method and system for a lithium iron phosphate power battery pack

Abstract

The present disclosure relates to a method and system for electromagnetic compatibility management of a lithium iron phosphate power battery pack. The multi-dimensional running data of the lithium iron phosphate power battery pack at multiple cycle times can be used to cover complex working conditions such as variable temperature, variable rate and dynamic SOC interval in actual vehicle operation, not only having a wider application range, but also being adaptable to lithium iron phosphate power battery packs of different capacities and different cell models. Based on the weighted correction of all multi-dimensional running data and combined with a deep learning model, the predicted radiation intensity is obtained. Compared with the existing PID algorithm or traditional neural network, the prediction accuracy and timing adaptability can be effectively improved, and the electromagnetic compatibility risk can be predicted in advance. Furthermore, based on the actual radiation intensity and the predicted radiation intensity, the parameters of the deep learning model are adjusted to dynamically optimize the parameters through error analysis of the closed-loop feedback mechanism, thereby realizing adaptive management.

Description

Technical Field

[0001] The embodiments in this specification belong to the field of power battery testing technology, and specifically relate to an electromagnetic compatibility control method and system for lithium iron phosphate power battery packs. Background Technology

[0002] With the rapid development of the new energy vehicle industry, lithium iron phosphate (LFP) power batteries have become the mainstream power source for passenger and commercial vehicles due to their high safety, long cycle life, and significant cost advantages. However, during the charging and discharging process, LFP power battery packs generate electromagnetic radiation ranging from 30MHz to 1GHz due to factors such as cell polarization, high-voltage circuit switching, and communication interference from the battery management system (BMS). If the electromagnetic compatibility (EMC) performance is substandard, it will interfere with the normal operation of the vehicle's electronic control system (such as autonomous driving sensors or onboard communication modules) and may even cause safety hazards.

[0003] Currently, existing electromagnetic compatibility control methods for power battery packs mainly adopt passive protection measures, such as adding filter devices with fixed parameters in the high-voltage circuit, using metal shields to encapsulate the battery pack, or optimizing cable routing. These methods have core defects such as lack of dynamic applicability, slow response and insufficient accuracy, failure to combine the characteristics of lithium iron phosphate and insufficient multi-module coordination, thus failing to ensure the normal operation of the vehicle's electronic control system. Summary of the Invention

[0004] The embodiments of this disclosure propose an electromagnetic compatibility control method and system for lithium iron phosphate power battery packs.

[0005] In a first aspect of this disclosure, an electromagnetic compatibility (EMC) control method for a lithium iron phosphate (LFP) power battery pack is provided. The method includes preprocessing multi-dimensional operational data of the LFP power battery pack across multiple cycles, where each multi-dimensional operational data point has multiple electromagnetic radiation intensities and multiple feature types. The method further includes weighting and correcting all preprocessed multi-dimensional operational data, and determining the predicted radiation intensity of the LFP power battery pack at a target time based on the weighted and corrected multi-dimensional operational data and a deep learning model. The method also includes determining a control strategy based on the predicted radiation intensity and a preset threshold range, and performing EMC control on the LFP power battery pack based on the control strategy. Furthermore, the method includes determining the actual radiation intensity of the LFP power battery pack at the target time, and adjusting the parameters of the deep learning model based on the actual radiation intensity and the predicted radiation intensity.

[0006] In a second aspect of this disclosure, an electromagnetic compatibility (EMC) control system for a lithium iron phosphate (LFP) power battery pack is provided. The system includes a data sensing module configured to preprocess multi-dimensional operational data of the LFP power battery pack across multiple cycles, where each multi-dimensional operational data point has multiple electromagnetic radiation intensities and multiple feature types. The system also includes a radiation intensity prediction module configured to weight and correct all preprocessed multi-dimensional operational data, and based on the weighted and corrected multi-dimensional operational data and a deep learning model, determine the predicted radiation intensity of the LFP power battery pack at a target time. The system further includes a control execution module configured to determine a control strategy based on the predicted radiation intensity and a preset threshold range, and to perform electromagnetic control on the LFP power battery pack based on the control strategy. Additionally, the system includes a parameter feedback module configured to determine the actual radiation intensity of the LFP power battery pack at the target time, and to adjust the parameters of the deep learning model based on the actual radiation intensity and the predicted radiation intensity.

[0007] In a third aspect of this disclosure, a computer program product is provided, comprising a computer program that is executed by a processor to implement the method according to the first aspect.

[0008] In a fourth aspect of this disclosure, a machine-readable storage medium is provided. The machine-readable storage medium stores machine-executable instructions, which are executed by a processor to implement the method provided according to a first aspect of this disclosure.

[0009] It should be understood that the description in the Summary of the Invention section is not intended to limit the key or essential features of the embodiments of this disclosure, nor is it intended to restrict the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0010] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. In the drawings, the same or similar reference numerals denote the same or similar elements, wherein:

[0011] Figure 1 A schematic diagram of an example environment in which some embodiments of this disclosure may be implemented is shown;

[0012] Figure 2 A flowchart illustrating an electromagnetic compatibility control method for a lithium iron phosphate power battery pack according to some embodiments of this disclosure is shown.

[0013] Figure 3 This diagram illustrates the overall process of electromagnetic compatibility control for a lithium iron phosphate power battery pack according to some embodiments of the present disclosure.

[0014] Figure 4 A block diagram of an electromagnetic compatibility control system for a lithium iron phosphate power battery pack, according to some embodiments of this disclosure, is shown; and

[0015] Figure 5 A block diagram of an electronic device that can implement several embodiments of the present disclosure is shown. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0017] The terms “comprising” and “having”, and any variations thereof, in this specification, claims, and the foregoing drawings are intended to cover a non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the steps or units listed, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to such process, method, product, or apparatus. Depending on the context, the word “if” as it applies herein may be interpreted as “when”, “when”, “in response to determination”, or “in response to detection”.

[0018] As mentioned above, existing electromagnetic compatibility (EMC) management methods for power battery packs mainly employ passive protection measures, such as adding filter devices with fixed parameters to the high-voltage circuit, using metal shielding covers to encapsulate the battery pack, or optimizing cable routing. These methods have the following core drawbacks:

[0019] Lack of dynamic applicability: The electromagnetic radiation intensity of lithium iron phosphate power batteries is strongly correlated with state parameters such as SOC, temperature, cycle number, and charge / discharge rate (e.g., when SOC=100%, the electromagnetic radiation intensity is 30%-50% higher than when SOC=20%, and the radiation spectrum shifts significantly in low-temperature environments). Passive protection with fixed parameters cannot match the electromagnetic compatibility variation under all operating conditions.

[0020] Lagging response and insufficient accuracy: Existing electromagnetic compatibility control methods for power battery packs are mostly based on PID algorithms or traditional neural networks, which can only provide feedback adjustment for electromagnetic compatibility exceedances that have already occurred. They not only lack the ability to predict in advance, but also have low fitting accuracy for electromagnetic radiation data with strong time series (prediction error generally exceeds ±8%, and response time is greater than 50ms).

[0021] The existing electromagnetic compatibility management methods for power battery packs do not take into account the characteristics of lithium iron phosphate cells. They do not optimize the algorithms for the crystal structure stability, low-temperature self-discharge characteristics and voltage plateau characteristics of lithium iron phosphate cells, resulting in management strategies that are general but poorly targeted.

[0022] Insufficient multi-module coordination: Existing electromagnetic compatibility control methods for power battery packs mostly involve adjusting the filter or shielding modules individually, failing to achieve coordinated linkage between the battery management system, filter system, and shielding system, resulting in control redundancy or lack of control.

[0023] These core defects can easily lead to electromagnetic compatibility performance failing to meet standards in real time, thus failing to guarantee the normal operation of the vehicle's electronic control system.

[0024] To address this, embodiments of this disclosure propose an electromagnetic compatibility (EMC) control method for lithium iron phosphate (LFP) power battery packs. The method includes preprocessing multi-dimensional operational data from the LFP power battery pack across multiple cycles, where each multi-dimensional operational data point exhibits multiple electromagnetic radiation intensities and multiple feature types. The method further includes weighting and correcting all preprocessed multi-dimensional operational data, and determining the predicted radiation intensity of the LFP power battery pack at a target time based on the weighted and corrected multi-dimensional operational data and a deep learning model. The method also includes determining a control strategy based on the predicted radiation intensity and a preset threshold range, and performing EMC control on the LFP power battery pack based on the control strategy. Furthermore, the method includes determining the actual radiation intensity of the LFP power battery pack at the target time, and adjusting the parameters of the deep learning model based on the actual radiation intensity and the predicted radiation intensity.

[0025] In this way, multi-dimensional operational data of lithium iron phosphate (LFP) battery packs across multiple cycles can be used to predict radiation intensity that covers complex operating conditions such as temperature variations, rate of change, and dynamic state of charge (SOC) ranges in actual vehicle operation. This not only broadens the application range but also adapts to LFP battery packs of different capacities and cell models. By weighting and correcting all multi-dimensional operational data and combining it with a deep learning model to obtain the predicted radiation intensity, the prediction accuracy and time-series adaptability can be effectively improved compared to existing PID algorithms or traditional neural networks, and electromagnetic compatibility risks can be predicted in advance. Furthermore, the parameters of the deep learning model are adjusted based on the actual and predicted radiation intensities to dynamically optimize the parameters through error analysis using a closed-loop feedback mechanism, thereby achieving adaptive control.

[0026] Figure 1 Schematic diagrams are shown illustrating example environments in which some embodiments of this disclosure can be implemented. For example... Figure 1As shown, the example environment 100 may include a data sensing device 101, which is used to collect multi-dimensional operating data of the lithium iron phosphate power battery pack under multiple cycles (i.e., charge and discharge cycles). Here, the data sensing device 101 specifically includes an electromagnetic radiation sensing unit, a battery status sensing unit, and an environmental sensing unit. The electromagnetic radiation sensing unit can collect the electromagnetic radiation intensity of the lithium iron phosphate battery pack over multiple cycles using multiple broadband electromagnetic sensors. For example, it can use four broadband electromagnetic sensors located at the four corners inside the lithium iron phosphate battery pack and a broadband electromagnetic sensor located near the high-voltage busbar to collect a total of five electromagnetic radiation intensities. These five intensities can be processed using weighted summation to obtain the corresponding effective radiation intensity. The battery status sensing unit can collect core parameters of the lithium iron phosphate battery pack over multiple cycles, such as SOC, individual cell temperature, charge / discharge rate, cycle count, and individual cell voltage, through the battery management system. The environmental sensing unit can collect the external ambient temperature of the lithium iron phosphate battery pack over multiple cycles using a temperature sensor and the external ambient humidity of the lithium iron phosphate battery pack over multiple cycles using a humidity sensor.

[0027] It is understandable that the broadband electromagnetic sensor, battery management system, temperature sensor and humidity sensor mentioned above are all well-known and mature battery testing devices in the field, and their hardware structure and testing process will not be elaborated here.

[0028] Example environment 100 may further include a processing terminal 102, which establishes a communication connection with data sensing device 101 to acquire multi-dimensional operating data of the lithium iron phosphate power battery pack over multiple cycles via CAN FD bus. This multi-dimensional operating data may include all the aforementioned electromagnetic radiation intensity, SOC, single-cell temperature, charge / discharge rate, cycle number, single-cell voltage, external ambient temperature, and external ambient humidity. All multi-dimensional operating data is preprocessed. Here, preprocessing methods may include normalization, outlier removal, and linear interpolation to effectively ensure the reliability and validity of all multi-dimensional operating data.

[0029] The processing terminal 102 can also perform weighted correction on all preprocessed multi-dimensional operational data, and determine the predicted radiation intensity of the lithium iron phosphate power battery pack at the target time based on all weighted and corrected multi-dimensional operational data and the deep learning model. Here, multiple core feature types that have a significant impact on the electromagnetic radiation of the lithium iron phosphate power battery pack can be screened based on all feature types of each multi-dimensional operational data, and the operational data corresponding to all core feature types can be weighted and corrected based on correction factors (the operational data corresponding to all remaining feature types do not need to be weighted and corrected).

[0030] It is understandable that the target time is a future time after the current time. By inputting all the weighted and corrected multi-dimensional running data into the pre-trained deep learning model, the predicted radiation intensity at the future time can be determined. This not only effectively improves the prediction accuracy and temporal adaptability, but also allows for the early prediction of electromagnetic compatibility risks.

[0031] Furthermore, the processing terminal 102 can also determine control strategies based on predicted radiation intensity and preset threshold intervals. Here, the preset threshold intervals have multiple threshold intervals, such as a first threshold interval, a second threshold interval, and a third threshold interval. The maximum value of the first threshold interval is less than the minimum value of the second threshold interval, and the maximum value of the second threshold interval is less than the minimum value of the third threshold interval. Each threshold interval has a corresponding risk level, and the control strategy corresponding to each risk level can be queried through the control strategy database.

[0032] Example environment 100 may also include electromagnetic control device 103, which establishes a communication connection with processing terminal 102 to obtain control strategy and perform electromagnetic control on lithium iron phosphate power battery pack based on the control strategy. Here, the electromagnetic control device 103 specifically includes an intelligent filtering device, an active shielding device, and a battery management coordination device. The intelligent filtering device can switch filtering parameters based on the control strategy to achieve precise filtering of electromagnetic interference in different frequency bands. It mainly consists of an FPGA controller, a reconfigurable capacitor array (adjustable range 1nF-10nF), a reconfigurable inductor array (adjustable range 1μH-10μH), and a filter topology switching circuit. The FPGA controller is used to drive the reconfigurable capacitor array to adjust capacitor parameters and / or drive the reconfigurable inductor array to adjust inductor parameters based on the control strategy. The reconfigurable capacitor array has multiple capacitor cells with nominal capacitance values, and each capacitor cell has a corresponding controllable switch. The reconfigurable inductor array has multiple inductor cells with nominal inductance values, and each inductor cell has a corresponding controllable switch. The filter topology switching circuit is used to switch the access topology of capacitors and / or inductors in the filter circuit.

[0033] The active shielding device can adjust the shielding layer driving voltage based on a control strategy and dynamically adjust the shielding effectiveness by switching the grounding method. It mainly consists of a flexible conductive shielding layer, an electromagnetic drive unit, a grounding switching switch, and a grounding method control circuit. The flexible conductive shielding layer can be a flexible conductive material such as conductive fabric, metal foil, or conductive coating, which covers the inner wall of the battery pack to form a complete electromagnetic shielding layer. The electromagnetic drive unit is used to provide an adjustable driving voltage (adjustable range set to 0-12V) to the flexible conductive shielding layer based on the control strategy to change the conductivity and shielding effectiveness of the flexible conductive shielding layer. The grounding switching switch has a multi-way switch to connect the flexible conductive shielding layer to the battery pack ground. The grounding method control circuit is used by the electromagnetic drive unit to switch the on / off state of the grounding switching switch based on the control strategy to realize the switching of the grounding method (such as single-point grounding or multi-point grounding).

[0034] The battery management coordination device establishes a communication connection with the battery management system (BMS) to adjust the charge and discharge rates of the lithium iron phosphate (LFP) power battery pack based on control strategies, thereby reducing electromagnetic radiation caused by differences in cell consistency. It mainly consists of a BMS communication interface, a command parsing and conversion unit, and a status monitoring unit. The BMS communication interface has a standardized protocol for communicating with the BMS to send control requests (such as adjusting the upper limit of the charge and discharge rate, with the adjustment range set between 0.1C and 3C) or to receive operating data fed back by the BMS. The command parsing and conversion unit is used to convert control commands into control commands recognizable by the BMS based on the control strategies. The status monitoring unit is used to monitor the operating data fed back by the BMS in real time.

[0035] It should be noted that the hardware structure and control principle of the intelligent filtering device, active shielding device and battery management coordination device mentioned above are all well-known technologies in this field, and will not be elaborated further here.

[0036] Furthermore, after the lithium iron phosphate power battery pack is electromagnetically controlled by the electromagnetic control device 103, the processing terminal 102 can determine the actual radiation intensity of the lithium iron phosphate power battery pack at a target time, and adjust the parameters of the deep learning model based on the actual radiation intensity and the predicted radiation intensity. Here, the actual radiation intensity can be determined based on multiple electromagnetic radiation intensities collected by the data sensing device 101 at the target time. By using the relative error obtained based on the actual radiation intensity and the predicted radiation intensity, the parameters of the aforementioned deep learning model can be adjusted when the relative error is within a preset error range.

[0037] It should be understood that the architecture and functionality in example environment 100 are described for illustrative purposes only and do not imply any limitation on the scope of this disclosure. Embodiments of this disclosure can also be applied to other environments with different architectures and / or functionalities.

[0038] Figure 2 This document illustrates a flowchart of an electromagnetic compatibility (EMC) control method for a lithium iron phosphate power battery pack, representing some embodiments of this disclosure. Method 200 may, for example, be derived from... Figure 1 The example environment shown executes a processing terminal, which can be deployed on an in-vehicle BMS or a cloud management platform. For example... Figure 2 As shown in box 202, method 200 can preprocess all multi-dimensional operational data based on the multi-dimensional operational data of the lithium iron phosphate power battery pack at multiple cycle counts. Here, the processing terminal can... Figure 1 The data sensing device shown is used to acquire multi-dimensional operating data of lithium iron phosphate power battery packs under multiple cycles. Each multi-dimensional operating data has multiple electromagnetic radiation intensities and multiple characteristic types of operating data. The multiple electromagnetic radiation intensities can be the electromagnetic radiation intensities at the four corners inside the lithium iron phosphate power battery pack and the electromagnetic radiation intensities near the high-voltage busbar. The multiple characteristic types of operating data can be SOC, single cell temperature, charge / discharge rate, number of cycles, single cell voltage, external ambient temperature, and external ambient humidity.

[0039] In some implementations, when preprocessing all multi-dimensional operational data, the processing terminal specifically performs a weighted summation of all electromagnetic radiation intensities in each multi-dimensional operational data set to obtain operational data with electromagnetic radiation intensity as the characteristic type. Here, taking the electromagnetic radiation intensities at the four corners inside the lithium iron phosphate battery pack and near the high-voltage busbar as examples, the calculation formula shown below can be used to determine the operational data with electromagnetic radiation intensity as the characteristic type:

[0040]

[0041] In the above formula, The data is operational data with electromagnetic radiation intensity as its characteristic type. , , , and The preset weighting coefficients, This refers to the electromagnetic radiation intensity near the high-voltage busbar. , , and The electromagnetic radiation intensity refers to the intensity at the four corners inside the lithium iron phosphate power battery pack.

[0042] Understandably, since the area near the high-voltage busbar is the core source of radiation, a preset weighting coefficient of 0.4 can be set for the electromagnetic radiation intensity near the high-voltage busbar, and preset weighting coefficients of 0.15 can be set for the electromagnetic radiation intensity at the four corners inside the lithium iron phosphate power battery pack. In addition, before weighting and summing all electromagnetic radiation intensities, outlier removal and normalization can be performed on all electromagnetic radiation intensities based on the Raida criterion, and this is not limited to this.

[0043] Next, based on all the multi-dimensional operational data, the operational datasets corresponding to each feature type are determined, and each operational dataset is normalized. Here, all feature types may include electromagnetic radiation intensity, and the aforementioned SOC, cell temperature, charge / discharge rate, cycle count, cell voltage, ambient temperature, and ambient humidity.

[0044] It is understandable that the running dataset corresponding to each feature type may include running data of all corresponding feature types corresponding to all loop counts, and can be arranged in order of loop count; the running dataset is normalized by using the Z-score standardization formula so that the normalized running data meets the standard normal distribution with a mean of 0 and a standard deviation of 1.

[0045] Next, the mean and standard deviation are determined based on the normalized running datasets, and multiple residuals are determined based on each mean and corresponding standard deviation. For example, the normalized running datasets are represented as follows: The mean, standard deviation, and residuals are obtained using the formulas shown below:

[0046]

[0047]

[0048]

[0049] In the above formula, Let n be the mean, and n be the number of data points in the dataset. Standard deviation, Let C be the i-th running data, and let C be the residual corresponding to the i-th running data.

[0050] Next, critical values ​​are determined based on the number of data points in each running dataset and the preset significance level parameter. Then, based on each critical value and all corresponding residuals, outlier removal is performed on the normalized running dataset. Here, the corresponding critical values ​​(i.e., the critical values ​​of the Grubbs statistic G) can be directly read from the standardized Grubbs critical value table based on the number of data points in each running dataset and the preset significance level parameter (e.g., set to 0.05). For example, when the number of data points is 40 and the preset significance level parameter is 0.05, the corresponding critical value can be directly read as 2.914. Furthermore, when the number of data points exceeds 100, the corresponding critical values ​​can also be obtained using the calculation formula shown below:

[0051]

[0052] In the above formula, This is the critical value when the preset significance level parameter is 0.05, and n is the number of data points.

[0053] Understandably, the maximum residual can be determined based on all residuals, and the Grubbs statistic corresponding to this maximum residual can be determined using the formula shown below:

[0054]

[0055] In the above, G represents the Grubbs statistic. For the maximum residual, The mean mentioned above, The standard deviation mentioned above.

[0056] When the Grubbs statistic corresponding to the maximum residual exceeds the critical value, the running data corresponding to the maximum residual can be identified as an outlier and outlier removal can be performed. After outlier removal, the maximum residual and the corresponding Grubbs statistic can be re-determined by referring to the above, and compared with the critical value again, until the finally determined Grubbs statistic does not exceed the critical value.

[0057] Next, linear interpolation is performed on each running dataset after outlier removal. It is understandable that, since the removed outliers are single or a few discrete data points in the running dataset, linear interpolation can be used to interpolate them. For example, the mean between two data points adjacent to the outlier is inserted into the position of the removed outlier in the running dataset.

[0058] In box 204, method 200 can perform weighted correction on all preprocessed multi-dimensional operational data, and determine the predicted radiation intensity of the lithium iron phosphate power battery pack at the target time based on all weighted corrected multi-dimensional operational data and the deep learning model. Here, the processing terminal can filter out several core feature types that have a significant impact on the electromagnetic radiation of the lithium iron phosphate power battery pack based on all feature types of each multi-dimensional operational data, and perform weighted correction on the operational data corresponding to all core feature types based on correction factors (the operational data corresponding to all remaining feature types do not need to be weighted).

[0059] In some implementations, when the processing terminal performs weighted correction on all the preprocessed multi-dimensional operational data, it specifically selects multiple core feature types from all feature types and determines the corresponding Pearson correlation coefficient based on the operational datasets of each core feature type and the operational datasets with electromagnetic radiation intensity as the feature type. Here, by combining the electromagnetic compatibility characteristics of the lithium iron phosphate power battery pack, SOC, single cell temperature, charge / discharge rate, and cycle number can be selected as multiple core feature types from all feature types, and the Pearson correlation coefficient corresponding to each core feature type is determined by referring to the calculation formula shown below:

[0060]

[0061] In the above formula, The Pearson correlation coefficient is the value corresponding to the i-th core feature type, and n is the number of data points in the dataset. For the j-th value in the running dataset of the i-th core feature type, The mean of the running dataset for the i-th core feature type. For the j-th value in the running dataset with the feature type of electromagnetic radiation intensity, This is the mean of the running dataset with electromagnetic radiation intensity as the feature type.

[0062] Next, correction factors for each core feature type are determined based on all Pearson correlation coefficients, and the running datasets for the corresponding core feature types are then weighted and corrected based on these correction factors. Here, taking four core feature types—SOC, single-cell temperature, charge / discharge rate, and cycle count—as an example, the correction factors for each core feature type can be determined using the calculation formulas shown below:

[0063]

[0064] In the above formula, For the correction factor of the i-th core feature type, The Pearson correlation coefficient is the coefficient corresponding to the i-th core feature type. The Pearson correlation coefficient is the coefficient corresponding to the first core feature type. The Pearson correlation coefficient corresponding to the second core feature type. The Pearson correlation coefficient is the coefficient corresponding to the third core feature type. The Pearson correlation coefficient is the coefficient corresponding to the fourth core feature type.

[0065] Of course, embodiments of this disclosure can also be used to simulate and test the correction factors for all core feature types based on the Long Short-Term Memory (LSTM) network to determine whether all correction factors need to be adjusted. For example, taking a training set with multiple historical data sets and corresponding radiation intensity labels for each training set as an example, each training set includes at least historical running data for all core feature types and historical running data for other feature types. The historical running data for all core feature types in each training set can be corrected based on all the correction factors mentioned above (e.g., by multiplying the historical running data for each core feature type with the corresponding correction factor), and the LTM network can be trained based on the corrected training sets and corresponding radiation intensity labels. After training the LTM network, the LTM network can be determined based on multiple validation sets and corresponding radiation intensity labels for each validation set. The long short-term memory network (LSM) can be fine-tuned after training based on all correction factors. If the average prediction error exceeds a preset error threshold (e.g., 3%), all correction factors can be fine-tuned (e.g., increasing correction factors for core feature types with large errors). After fine-tuning all correction factors, the LSM network can be retrained as described above. The average prediction error of the LSM network after retraining can be determined to be greater than the preset error threshold until the average prediction error of the LSM network after training is less than or equal to the preset error threshold. All correction factors used in the last training of the LSM network are then determined as the final set of correction factors.

[0066] Next, after determining the correction factors for all core feature types, the running datasets for the corresponding core feature types can be weighted and corrected based on each correction factor, as shown in the calculation formula below:

[0067]

[0068] In the above formula, The i-th value after weighted correction in the running dataset for each core feature type. For each core feature type, the i-th value in the running dataset. These are the correction factors for each core feature type. It is understandable that when the weighted corrected value in the running dataset for each core feature type is greater than 1, further normalization is unnecessary.

[0069] In some implementations, when the processing terminal determines the predicted radiation intensity of the lithium iron phosphate power battery pack at a target time based on all weighted and corrected multi-dimensional operational data and a deep learning model, it specifically determines the average historical radiation intensity corresponding to each cycle number from all the weighted and corrected multi-dimensional operational data, along with all operational data with electromagnetic radiation intensity as the feature type and the historical radiation intensity set. It is understood that using the average historical radiation intensity as an input feature of the deep learning model can provide a more direct reference to recent trends and reduce the interference of instantaneous noise on the prediction results, thereby making the prediction of the deep learning model more stable.

[0070] Here, taking a time window of 10 cycles as an example, the average historical radiation intensity corresponding to the first cycle among all the cycles mentioned above can be the average electromagnetic radiation intensity of the 10 historical cycles closest to that cycle. The average historical radiation intensity corresponding to the second cycle among all the cycles can be the average of the electromagnetic radiation intensity of all the 9 historical cycles closest to that cycle and the electromagnetic radiation intensity of the first cycle. And so on, to determine the average historical radiation intensity corresponding to each cycle among all the cycles.

[0071] Subsequently, the weighted and corrected multi-dimensional operational data and the average historical radiation intensity are input into the deep learning model to obtain the predicted radiation intensity of the lithium iron phosphate power battery pack at the target time. Here, based on the weighted and corrected multi-dimensional operational data and the average historical radiation intensity, a feature vector set represented by a multi-dimensional vector can be integrated. The dimension of this vector corresponds to all feature types and the average historical radiation intensity included in each multi-dimensional operational data. For example, taking the aforementioned multi-dimensional operational data including electromagnetic radiation intensity, SOC, single cell temperature, charge / discharge rate, cycle count, single cell voltage, external ambient temperature, and external ambient humidity as an example, the feature vector set has a nine-dimensional vector, corresponding to electromagnetic radiation intensity, SOC, single cell temperature, charge / discharge rate, cycle count, single cell voltage, external ambient temperature, external ambient humidity, and the average historical radiation intensity, respectively.

[0072] Understandably, deep learning models can be constructed using bidirectional long short-term memory networks (Bi-LSTM). The system consists of a bidirectional long short-term memory (LSTM) network layer, an attention mechanism (Transformer) layer, and an output layer. The bidirectional LSTM network layer has multiple input neurons corresponding to a feature vector set represented by a multi-dimensional vector. For example, when the feature vector set is a nine-dimensional vector as mentioned above, the bidirectional LSTM network layer can have nine input neurons. It obtains the hidden state sequence corresponding to the feature vector set through a forward LSTM network, a backward LSTM network, and hidden layers. The attention mechanism layer takes the hidden state sequence output by the bidirectional LSTM network layer as input, calculates the query, key, and value matrices through linear transformations, and determines multiple attention weight coefficients corresponding to the feature vector set based on these matrices (e.g., calculating the attention score matrix and the context vector corresponding to the attention score matrix sequentially, and mapping the context vector to multiple attention weight coefficients). For example, when the feature vector set is a nine-dimensional vector as mentioned above, the multiple attention weight coefficients can be a nine-dimensional vector composed of nine attention weight coefficients. The output layer can be a fully connected network, which calculates the predicted radiation intensity by operating on the multiple attention weight coefficients output by the attention mechanism layer and the feature vector set.

[0073] In box 206, method 200 can determine a control strategy based on the predicted radiation intensity and a preset threshold range, and perform electromagnetic control on the lithium iron phosphate power battery pack based on the control strategy. Here, the preset threshold range has multiple threshold ranges, such as a first threshold range, a second threshold range, and a third threshold range. The maximum value of the first threshold range is less than the minimum value of the second threshold range, and the maximum value of the second threshold range is less than the minimum value of the third threshold range. Each threshold range has a corresponding risk level, and the control strategy corresponding to each risk level can be retrieved through a control strategy library.

[0074] In some implementations, a preset threshold range is divided into a first threshold range, a second threshold range, and a third threshold range. For example, the first threshold range is (0, 30 dBμV / m), the second threshold range is (30 dBμV / m, 40 dBμV / m), and the third threshold range is greater than 40 dBμV / m. When the processing terminal determines a control strategy based on the predicted radiation intensity and the preset threshold range, specifically based on the predicted radiation intensity being within the first threshold range, it determines a control strategy corresponding to the low-risk level from the control strategy library. This control strategy is used to switch filtering parameters (such as electrical...). Alternatively, based on the predicted radiation intensity being in the second threshold range, a control strategy corresponding to the medium-risk level is determined in the control strategy library. This control strategy is used to switch filter parameters (such as inductance parameters) and switch grounding methods. Or, based on the predicted radiation intensity being in the third threshold range, a control strategy corresponding to the high-risk level is determined in the control strategy library. This control strategy is used to switch filter parameters (such as capacitance and inductance parameters), adjust the shielding layer driving voltage, and adjust the charge and discharge rate. It can also activate the battery pack cooling system to reduce the cell temperature so that the cell temperature is below 45°C.

[0075] Understandably, after determining the control strategy, the processing terminal can... Figure 1 The electromagnetic control device shown feeds back the control strategy, enabling it to perform electromagnetic control on the lithium iron phosphate battery pack based on this strategy. Specifically, the electromagnetic control device includes an intelligent filtering device, an active shielding device, and a battery management coordination device. The intelligent filtering device can switch filtering parameters (such as capacitor parameters and / or inductor parameters) based on the control strategy to achieve precise filtering of electromagnetic interference in different frequency bands. It mainly consists of an FPGA controller, a reconfigurable capacitor array, a reconfigurable inductor array, and a filter topology switching circuit. The active shielding device can adjust the shielding layer driving voltage based on the control strategy and / or dynamically adjust the shielding effectiveness by switching the grounding method. It mainly consists of a flexible conductive shielding layer, an electromagnetic drive unit, a grounding switching switch, and a grounding method control circuit. The battery management coordination device establishes a communication connection with the battery management system to adjust the charge / discharge rate of the lithium iron phosphate battery pack based on the control strategy to reduce electromagnetic radiation caused by differences in cell consistency. It mainly consists of a BMS communication interface, an instruction parsing and conversion unit, and a status monitoring unit.

[0076] In some implementations, when the processing terminal performs electromagnetic control on the lithium iron phosphate power battery pack based on the control strategy, it specifically controls an intelligent filtering device to perform electromagnetic control on the lithium iron phosphate power battery pack based on the control strategy corresponding to the low-risk level. The intelligent filtering device is used to adjust the capacitor parameters based on the control strategy. For example, when the control strategy is a precise capacitance value corresponding to the current electromagnetic radiation frequency band, the intelligent filter can directly switch the nominal capacitance value of a single capacitor in the reconfigurable capacitor array to match the precise capacitance value corresponding to the electromagnetic radiation frequency band, and connect it to the filter topology switching circuit, while keeping the active shielding device in a basic state (e.g., controlling the shielding layer driving voltage as the base voltage); or, when the control strategy is a precise capacitance value corresponding to the current electromagnetic radiation frequency band, and does not match the capacitance value of any single capacitor, the intelligent filter can also combine multiple capacitors in the reconfigurable capacitor array in series and parallel (e.g., connecting two 5nF capacitors in series to obtain a 2.5nF equivalent capacitance, or connecting three 3nF capacitors in parallel to obtain a 9nF equivalent capacitance) to match the precise capacitance value corresponding to the electromagnetic radiation frequency band, and connect it to the filter topology switching circuit; or, when the control strategy is a specific risk level, the intelligent filter can control the capacitance value connected to the filter topology switching circuit in the reconfigurable capacitor array to be within the capacitance range of the corresponding risk level (e.g., low risk level). The corresponding capacitance range can be 1-3nF, the capacitance range for medium-risk levels can be 4-7nF, and the capacitance range for high-risk levels can be 8-10nF; or, when the control strategy is a specific growth trend of capacitance value with the increase of SOC, the intelligent filter device can control the capacitance value connected to the filter topology switching circuit in the reconfigurable capacitor array to gradually increase with the increase of SOC (e.g., during the process of SOC increasing from 80% to 100%, the capacitance value can be continuously and continuously finely adjusted from 4nF to 7nF in real time), so as to adapt to the dynamic change law of electromagnetic radiation through millisecond-level continuous fine-tuning of capacitance value; or, when the control strategy is a specific external ambient temperature or cycle number, the intelligent filter device can control the capacitance value connected to the filter topology switching circuit in the reconfigurable capacitor array to correspond to the external ambient temperature (e.g., when the external ambient temperature is -10℃, the corresponding capacitance value can be adjusted to 6nF), or to correspond to the cycle number (e.g., when the cycle number exceeds 2000 times, the corresponding capacitance value can be adjusted to 9nF).

[0077] Furthermore, based on the control strategy corresponding to the medium-risk level, the active shielding device and the intelligent filtering device are used to control the electromagnetic interference of the lithium iron phosphate battery pack. The active shielding device is used to adjust the grounding method based on the control strategy, and the intelligent filtering device is used to adjust the inductor parameters based on the control strategy. For example, when the control strategy has a specific grounding method (such as single-point grounding), the active shielding device can control the on / off state of the switching switch to switch the grounding method of the grounding method control circuit to single-point grounding; and when the control strategy has a precise inductance value corresponding to the current electromagnetic radiation frequency band (such as 5μH when the current electromagnetic radiation frequency band is 600MHz under the medium-risk level), the intelligent filtering device can directly switch the nominal inductance value of a single inductor in the reconfigurable inductor array to match the precise inductance value corresponding to the electromagnetic radiation frequency band and connect it to the filter topology switching circuit; or, when the control strategy has a precise inductance value corresponding to the current electromagnetic radiation frequency band... When the corresponding precise inductance value does not match the inductance value of any single inductor, the intelligent filter can also combine multiple inductors in the reconfigurable inductor array in series and parallel (e.g., connecting a 5μH and a 2μH inductor in series to obtain an equivalent 7μH inductor) to match the precise inductance value corresponding to the electromagnetic radiation frequency band, and connect it to the filter topology switching circuit; or, when the control strategy has a specific core air gap, the intelligent filter can adjust the core structure with electromagnetic adjustment in the reconfigurable inductor array to match the corresponding core air gap, thereby changing the actual inductance value of a single inductor by adjusting the core air gap (e.g., reducing the core air gap of an 8μH inductor to make the actual inductance value of the single inductor less). The inductance value is slightly increased to 8.5μH, shortening the adjustment response time while adapting to spectrum changes; or, when the control strategy has a specific charge / discharge rate, the intelligent filter device can control the individual inductors connected to the filter topology switching circuit in the reconfigurable inductor array to switch to the inductor corresponding to the charge / discharge rate (e.g., when the charge / discharge rate is 2C, a 3μH inductor with a rated current of 5A can be switched to an inductor with the same inductance value and a rated current of 10A); or, when the control strategy has an inductor Q value corresponding to the electromagnetic radiation frequency band, the intelligent filter device can switch the individual inductors in the reconfigurable inductor array to individual inductors with the corresponding inductor Q value (e.g., needle). For electromagnetic radiation in the 30MHz-100MHz wideband, the high-Q ferrite core inductor is switched to a low-Q iron powder core inductor to achieve effective filtering across the wideband and cover low-voltage radiation interference from the battery pack. Alternatively, when the control strategy has a filtering topology access form corresponding to the risk level, the intelligent filtering device can control the filtering topology switching circuit to switch to the corresponding filtering topology access form (e.g., under medium-risk levels, the 4μH inductor is switched from being connected in series with the high-voltage bus filter circuit to being connected in parallel with ground to coordinate with the inductance value adjustment and enhance the suppression effect on pulsed electromagnetic radiation generated by the switching action of the power battery pack).

[0078] In addition, based on the control strategy corresponding to the high-risk level, the battery management coordination device, active shielding device and intelligent filtering device are used to electromagnetically control the lithium iron phosphate power battery pack. The battery management coordination device is used to adjust the charge and discharge rate based on the control strategy, the active shielding device is used to adjust the drive voltage based on the control strategy, and the intelligent filtering device is used to adjust the capacitor and inductor parameters based on the control strategy. For example, when the control strategy has a specific charge / discharge rate, the battery management coordination device can send a control command containing that charge / discharge rate to the battery management system; and when the control strategy has a specific drive voltage, the active shielding device can control the output of the corresponding drive voltage to the flexible conductive shielding layer; and when the control strategy has precise capacitance and precise inductance values ​​corresponding to the current electromagnetic radiation frequency band (e.g., when the current electromagnetic radiation frequency band is 600MHz under high-risk levels, the corresponding precise capacitance value can be 8nF and the inductance value can be 8μH), the intelligent filtering device can directly switch the nominal capacitance value of a single capacitor in the reconfigurable capacitor array and the nominal inductance value of a single inductor in the reconfigurable inductor array to match the precise capacitance and precise inductance values ​​corresponding to the electromagnetic radiation frequency band, and connect to the filter topology switching circuit; or, when the control strategy has a value corresponding to the SOC When calibrating the electromagnetic radiation frequency band and the current electromagnetic radiation frequency band, the intelligent filter device can keep the reconfigurable inductor array unchanged and fine-tune the individual capacitors connected to the filter topology switching circuit in the reconfigurable capacitor array (e.g., when the electromagnetic radiation spectrum shifts from 350MHz to 380MHz when SOC=100%, the inductance is fixed at 3μH, and the capacitor is fine-tuned from 3nF to 3.5nF); or, when the control strategy has capacitor combination and inductor combination methods corresponding to the risk level, the intelligent filter device can directly switch the capacitor combination method in the reconfigurable capacitor array and the inductor combination method in the reconfigurable inductor array (e.g., connecting 2 sets of 4nF capacitors in parallel (i.e., equivalent to 8nF) + 2 sets of 5μH inductors in series (i.e., equivalent to 10μH) to form a high capacitance and high inductance combination of 8nF+10μH, which is suitable for filtering strong electromagnetic radiation in the 1GHz high-frequency band).

[0079] Of course, in the embodiments of this disclosure, when the intelligent filtering device is used to adjust the capacitor and inductor parameters based on the control strategy, it can also utilize arrayed multi-combination time-sharing adjustment (e.g., when the battery pack is charging and discharging, it simultaneously generates 200MHz and 800MHz dual-band radiation, and the array time-sharing switches between two combinations of 2nF+2μH (200MHz) and 9nF+9μH (800MHz)), or step-by-step combination adjustment according to electromagnetic compatibility risk level (e.g., setting low risk to use 1nF+1μH light filtering combination, only...). Adjust the capacitance / single parameter; switch to the 4nF+4μH normal combination for medium risk, and adjust the capacitance and inductance synchronously; enable the 10nF+10μH strong filtering combination for high risk) and adjust the characteristic combination based on the battery status (e.g., when setting low temperature (-10℃) and SOC=20%, the battery radiation spectrum shifts and the intensity increases, matching the preset 5nF+6μH low temperature characteristic combination; when setting high temperature (45℃) and 1000 cycles, match the 7nF+5μH aging characteristic combination), and I will not go into too much detail here.

[0080] In box 208, method 200 can determine the actual radiation intensity of the lithium iron phosphate power battery pack at the target time, and adjust the parameters of the deep learning model based on the actual radiation intensity and the predicted radiation intensity. Here, the actual radiation intensity can be determined based on multiple electromagnetic radiation intensities collected by the aforementioned data sensing device at the target time (see above). The parameters of the aforementioned deep learning model are adjusted by using the relative error obtained based on the actual radiation intensity and the predicted radiation intensity, when the relative error is within a preset error range.

[0081] In some implementations, when the processing terminal adjusts the parameters of a deep learning model based on the actual and predicted radiation intensities, it specifically determines the relative error based on the actual and predicted radiation intensities, and determines whether the relative error is within a preset error range. Here, the relative error can be determined based on the actual and predicted radiation intensities, referring to the calculation formula shown below:

[0082]

[0083] In the above formula, This is a relative error. This represents the actual radiation intensity. To predict radiation intensity.

[0084] Understandably, the preset error range can be set to (3%, 5%). When the relative error is less than the preset error range, the current control strategy can be maintained. When the relative error is within the preset error range, the parameters of the aforementioned deep learning model can be adjusted. When the relative error is greater than the preset error range, the aforementioned deep learning model can be retrained (or the hidden layer parameters in the aforementioned deep learning model can be retrained). Furthermore, the control strategy library can be updated (such as redividing the threshold ranges corresponding to different risk levels and / or adjusting the parameters of multiple control strategies in the control strategy library) to ensure that the electromagnetic compatibility control method as a whole can adaptively optimize.

[0085] It should be noted that when the relative error is within a preset error range, a correction coefficient is determined based on the relative error and a preset scaling factor, and the parameters of the deep learning model are adjusted based on this correction coefficient. Here, the correction coefficient, based on the relative error and the preset scaling factor, can be determined using the calculation formula shown below:

[0086]

[0087] In the above formula, Here, k is the correction factor, and k is the preset proportionality factor. This represents the relative error.

[0088] Furthermore, the parameters that need to be adjusted in a deep learning model can be understood as the weight matrix obtained by the weight parameters of the output layer (i.e., the contribution of all feature types to the prediction result) after the deep learning model has been trained. The parameters of the deep learning model are adjusted by multiplying the sum of the aforementioned correction coefficients and 1 with the weight matrix.

[0089] Please see Figure 3 The diagram illustrates the overall electromagnetic compatibility control process of a lithium iron phosphate power battery pack according to some embodiments of this disclosure. Figure 3As shown, the overall electromagnetic compatibility management process 300 for lithium iron phosphate power battery packs includes preprocessing (i.e., data normalization + outlier removal + feature extraction) of multi-dimensional operational data of the lithium iron phosphate power battery pack under multiple cycles (i.e., radiation data collected by broadband electromagnetic sensors, BMS parameters collected by the battery management system, and external temperature and humidity of the battery pack collected by environmental sensors). A correction factor is then introduced to weight and correct multiple core feature types among all feature types. Next, the predicted radiation intensity of the lithium iron phosphate power battery pack at the target time is obtained based on a deep learning model. The system determines the threshold range within which the predicted radiation intensity falls. When the predicted radiation intensity is within the threshold range of 30 dBμV / m or less, a control strategy corresponding to the low-risk level can be determined and used to switch filter parameters (such as dynamically adjusting capacitor parameters). When the predicted radiation intensity is within the threshold range of 30 dBμV / m or greater than 40 dBμV / m, a control strategy corresponding to the medium-risk level can be determined and used to switch filter parameters (such as adjusting inductor parameters) and switch grounding methods. When the predicted radiation intensity is within the threshold range of 40 dBμV / m or greater, a control strategy corresponding to the high-risk level can be determined and used to switch filter parameters (such as adjusting inductor parameters) and switch grounding methods. The system implements corresponding control strategies based on the battery level, and is used to switch filter parameters (such as capacitor and inductor parameters), adjust the shielding layer drive voltage, and adjust the charge / discharge rate. It can also activate the battery pack cooling system to reduce cell temperature to below 45°C. Next, it determines the actual radiation intensity of the lithium iron phosphate battery pack at the target time, calculates the relative error based on the actual and predicted radiation intensities, and identifies the error range of the relative error. When the relative error is less than or equal to 3%, the current control strategy is maintained; when the relative error is greater than 3% but less than or equal to 5%, the strategy is adjusted accordingly. When the relative error is within the error range, the parameters of the aforementioned deep learning model can be adjusted; when the relative error is greater than 5%, the aforementioned deep learning model can be retrained (or the hidden layer parameters in the aforementioned deep learning model can be retrained), and the control strategy library can also be updated (such as re-dividing the threshold ranges corresponding to different risk levels, and / or adjusting the parameters of multiple control strategies in the control strategy library); then, multi-dimensional operating data of the lithium iron phosphate power battery pack for multiple cycles in the next cycle can be obtained, and the above can be referred to for continuous monitoring.

[0090] Figure 4A block diagram of an electromagnetic compatibility control system for a lithium iron phosphate power battery pack according to some embodiments of this disclosure is shown. The various embodiments in this specification are described in a progressive manner, with reference to each other for similar or identical parts. Each embodiment focuses on describing the differences from other embodiments. In particular, the device embodiments are basically similar to the method embodiments, so the description is relatively simple, and relevant parts can be referred to the description of the method embodiments. Figure 4 As shown, the electromagnetic compatibility (EMC) control system 400 for a lithium iron phosphate (LFP) power battery pack may include a data sensing module 402, configured to preprocess all multi-dimensional operating data based on the LFP power battery pack's multi-dimensional operating data over multiple cycles. Each multi-dimensional operating data point contains multiple electromagnetic radiation intensities and multiple feature types. The system 400 also includes a radiation intensity prediction module 404, configured to weight and correct all preprocessed multi-dimensional operating data, and determine the predicted radiation intensity of the LFP power battery pack at a target time based on the weighted and corrected multi-dimensional operating data and a deep learning model. Finally, the system 400 further includes a control execution module 406, configured to determine a control strategy based on the predicted radiation intensity and a preset threshold range, and perform electromagnetic control on the LFP power battery pack based on the control strategy. In addition, the electromagnetic compatibility control system 400 of the lithium iron phosphate power battery pack also includes a parameter feedback module 408, which is configured to determine the actual radiation intensity of the lithium iron phosphate power battery pack at a target time, and adjust the parameters of the deep learning model based on the actual radiation intensity and the predicted radiation intensity.

[0091] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this specification are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in or transmitted through a computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, Digital Subscriber Line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. The available media may be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., Digital Versatile Discs (DVDs)), or semiconductor media (e.g., Solid State Disks (SSDs)).

[0092] Figure 5 Block diagrams of electronic devices that can implement various embodiments of the present disclosure are shown. For example... Figure 5 As shown, the electronic device 500 includes a processor 501, which can perform various appropriate actions and processes based on computer program instructions loaded into random access memory (RAM) 503 according to computer program instructions stored in read-only memory (ROM) 502. The RAM 503 may also store various programs and data required for the operation of the electronic device 500. The processor 501, ROM 502, and RAM 503 are interconnected via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0093] The various processes and procedures described above, such as method 200, can be executed by processor 501. For example, in some embodiments, method 200 may be implemented as a software program tangibly contained in a machine-readable medium. In some embodiments, part or all of the software program may be loaded into and / or installed onto electronic device 500 via ROM 502. When the software program is loaded into RAM 503 and executed by processor 501, one or more actions of method 200 described above may be performed.

[0094] The functions described above in this document can be performed at least in part by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: field programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload programmable logic devices (CPLDs), and so on.

[0095] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0096] This disclosure can be a method, apparatus, system, and / or program product. The program product may include a machine-readable storage medium on which machine-readable program instructions for performing various aspects of this disclosure are loaded. The machine-readable program instructions described herein can be downloaded from the machine-readable storage medium to various computing / processing devices, or downloaded via a network, such as the Internet, a local area network, a wide area network, and / or a wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmissions, wireless transmissions, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the machine-readable program instructions from the network and forwards them to the machine-readable storage medium in the respective computing / processing device.

[0097] Machine program instructions used to perform the operations of this disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and conventional procedural programming languages ​​such as the "C" language or similar programming languages. Machine-readable program instructions may be executed entirely on a user's computer, partially on a user's computer, as a standalone software package, partially on a user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), is personalized by utilizing state information from the machine-readable program instructions to implement various aspects of this disclosure.

[0098] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing. Furthermore, although operations are depicted in a specific order, this should be understood as requiring that such operations be performed in the specific order shown or in sequential order, or requiring that all illustrated operations be performed to achieve the desired result. In certain environments, multitasking and parallel processing may be advantageous. Similarly, while several specific implementation details are included in the foregoing discussion, these should not be construed as limiting the scope of this disclosure. Certain features described in the context of individual embodiments may also be implemented in combination in a single implementation. Conversely, various features described in the context of a single implementation may also be implemented individually or in any suitable sub-combination in multiple implementations.

[0099] Although the subject matter has been described using language specific to structural features and / or methodological logic, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. Rather, the specific features and actions described above are merely illustrative examples of implementing the claims.

Claims

1. A method for electromagnetic compatibility control of a lithium iron phosphate power battery pack, characterized in that, include: Based on the multi-dimensional operation data of lithium iron phosphate power battery pack under multiple cycles, all the multi-dimensional operation data are preprocessed, and each of the multi-dimensional operation data has multiple electromagnetic radiation intensities and multiple characteristic types of operation data. All the preprocessed multi-dimensional operational data are weighted and corrected, and based on the weighted and corrected multi-dimensional operational data and the deep learning model, the predicted radiation intensity of the lithium iron phosphate power battery pack at the target time is determined. A control strategy is determined based on the predicted radiation intensity and a preset threshold range, and electromagnetic control is performed on the lithium iron phosphate power battery pack based on the control strategy. as well as The actual radiation intensity of the lithium iron phosphate power battery pack at the target time is determined, and the parameters of the deep learning model are adjusted based on the actual radiation intensity and the predicted radiation intensity.

2. The method according to claim 1, characterized in that, The preprocessing of all the multi-dimensional operational data includes: The electromagnetic radiation intensity of each of the multi-dimensional operational data is weighted and summed to obtain operational data with electromagnetic radiation intensity as the feature type. Based on all the multi-dimensional operational data, determine the operational dataset corresponding to each of the feature types, and normalize each of the operational datasets; The mean and standard deviation are determined based on each of the normalized running datasets, and multiple residuals are determined based on each mean and the corresponding standard deviation. A critical value is determined based on the number of data points in each of the aforementioned running datasets and a preset significance level parameter. Then, based on each critical value and all corresponding residuals, outlier removal processing is performed on the normalized running datasets. Linear interpolation is performed on each of the running datasets after outlier removal.

3. The method according to claim 2, characterized in that, The weighted correction of all the preprocessed multi-dimensional operational data includes: Multiple core feature types are selected from all the aforementioned feature types, and based on the running datasets for each of the core feature types and the running datasets for which the feature type is electromagnetic radiation intensity, the corresponding Pearson correlation coefficients are determined; and Based on all the Pearson correlation coefficients, a correction factor is determined for each of the core feature types, and the running datasets of the corresponding core feature types are weighted and corrected based on each correction factor.

4. The method according to claim 2, characterized in that, The determination of the predicted radiation intensity of the lithium iron phosphate power battery pack at the target time, based on all the weighted and corrected multi-dimensional operational data and the deep learning model, includes: Based on all the weighted and corrected multi-dimensional operational data, including all operational data with electromagnetic radiation intensity as the characteristic type and the historical radiation intensity set, the average historical radiation intensity corresponding to each cycle number is determined; and The weighted and corrected multi-dimensional operational data and the average historical radiation intensity are input into the deep learning model to obtain the predicted radiation intensity of the lithium iron phosphate power battery pack at the target time.

5. The method according to claim 1, characterized in that, The preset threshold intervals are divided into a first threshold interval, a second threshold interval, and a third threshold interval. The maximum value of the first threshold interval is less than the minimum value of the second threshold interval, and the maximum value of the second threshold interval is less than the minimum value of the third threshold interval. The determination of the control strategy based on the predicted radiation intensity and the preset threshold range includes: Based on the predicted radiation intensity falling within the first threshold range, a control strategy corresponding to the low-risk level is determined from the control strategy library; or Based on the predicted radiation intensity falling within the second threshold range, a control strategy corresponding to the medium-risk level is determined from the control strategy library; or Based on the predicted radiation intensity falling within the third threshold range, a control strategy corresponding to the high-risk level is determined from the control strategy library.

6. The method according to claim 5, characterized in that, The electromagnetic control of the lithium iron phosphate power battery pack based on the control strategy includes: Based on the control strategy corresponding to the low-risk level, an intelligent filtering device is used to perform electromagnetic control on the lithium iron phosphate power battery pack. The intelligent filtering device is used to adjust the capacitor parameters based on the control strategy; or Based on the control strategy corresponding to the medium-risk level, the active shielding device and the intelligent filtering device are used to perform electromagnetic control on the lithium iron phosphate power battery pack. The active shielding device is used to adjust the grounding method based on the control strategy, and the intelligent filtering device is used to adjust the inductance parameters based on the control strategy; or Based on the control strategy corresponding to the high-risk level, the battery management coordination device, the active shielding device, and the intelligent filtering device are used to perform electromagnetic control on the lithium iron phosphate power battery pack. The battery management coordination device is used to adjust the charge and discharge rate based on the control strategy, the active shielding device is used to adjust the drive voltage based on the control strategy, and the intelligent filtering device is used to adjust the capacitor and inductor parameters based on the control strategy.

7. The method according to claim 5, characterized in that, The step of adjusting the parameters of the deep learning model based on the actual radiation intensity and the predicted radiation intensity includes: The relative error is determined based on the actual radiation intensity and the predicted radiation intensity, and it is determined whether the relative error is within a preset error range; In response to determining that the relative error is within the preset error range, a correction coefficient is determined based on the relative error and a preset scaling factor, and the parameters of the deep learning model are adjusted based on the correction coefficient; or In response to determining that the relative error is greater than the maximum value of the preset error range, the deep learning model is retrained and the control strategy library is updated.

8. An electromagnetic compatibility control system for a lithium iron phosphate power battery pack, characterized in that, include: The data sensing module is configured to preprocess all the multi-dimensional operating data based on the multi-dimensional operating data of the lithium iron phosphate power battery pack under multiple cycles, and each of the multi-dimensional operating data has multiple electromagnetic radiation intensities and multiple feature types of operating data. The radiation intensity prediction module is configured to perform weighted correction on all the preprocessed multi-dimensional operating data, and determine the predicted radiation intensity of the lithium iron phosphate power battery pack at the target time based on all the weighted corrected multi-dimensional operating data and the deep learning model. The control execution module is configured to determine a control strategy based on the predicted radiation intensity and a preset threshold range, and to perform electromagnetic control on the lithium iron phosphate power battery pack based on the control strategy. as well as The parameter feedback module is configured to determine the actual radiation intensity of the lithium iron phosphate power battery pack at the target time, and to adjust the parameters of the deep learning model based on the actual radiation intensity and the predicted radiation intensity.

9. A computer-readable storage medium having a computer program stored thereon, the computer-readable storage medium storing instructions that, when executed on a computer or processor, cause the computer or processor to perform the steps of the method as claimed in any one of claims 1-7.

10. An electronic device, characterized in that, include: One or more processors, and A memory associated with the one or more processors, the memory being used to store program instructions that, when read and executed by the one or more processors, perform the steps of the method as described in any one of claims 1-7.

Images