End-cloud collaborative battery management method and system based on embedded optical fiber sensing
Through the collaborative work of embedded fiber sensing and cloud system, a distributed temperature self-perception and circuit model is built, and the SOC and thermal production rates are estimated and predicted by Kalman filters and LSTM networks are used to solve the internal information of the battery management system. The monitoring problem improves the accuracy and efficiency of battery management.
Patent Information
- Application Number
- CN202411113885.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-14
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2044-08-14
AI Technical Summary
Existing battery management systems are difficult to accurately monitor the internal information of lithium-ion batteries, traditional monitoring parameters are limited, intelligent embedded sensing technology is insufficiently applied, and there is a lack of solutions to improve the level of battery management system construction using cloud systems and digital twin technologies.
Using the end-cloud collaborative battery management method based on embedded fiber sensing, the internal temperature data of the battery is obtained through a distributed temperature self-sensing system, an equivalent circuit model and thermal collector model are built, and an extended Kalman filter and long-term memory network are used. Real-time estimation and prediction of SOC and thermal production rates are carried out to achieve multi-level fusion of data and online update of models.
It significantly improves the accuracy and efficiency of battery electric-thermal management, provides higher model accuracy and adaptability, and ensures richness of state estimation.
Smart Images

Figure CN119024174A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of battery management systems, and specifically relates to an end-cloud collaborative battery management method and system based on embedded optical fiber sensing. Background Art
[0002] In the practical application of lithium-ion batteries, real-time and accurate monitoring and management are the key to ensuring the safe and reliable operation of electrical equipment. At present, battery management still generally relies on the program pre-installed on the terminal to analyze the basic signals collected by the sampling circuit to realize the monitoring of the battery operation status. However, the battery system is a nonlinear system with strong time-varying and temperature-dependent characteristics, which often leads to many difficulties in accurately monitoring the key electrothermal behavior of the battery. In the framework of the traditional battery management system, the available monitoring parameters are limited to current, voltage and surface temperature, which focuses more on designing complex algorithms to improve management efficiency. However, since it is impossible to obtain more measurement data to more intuitively infer the internal information of the battery, the real effect of battery management is limited. In the recent existing battery management technology that has partially utilized intelligent embedded sensors, although it has achieved certain improvements compared with the traditional battery management system, the battery model used still has a large room for improvement, and the coordination and matching between the terminal system is not perfect enough. In addition, the system architecture integrating advanced sensing technology has a wider and richer amount of data than the traditional architecture. Those skilled in the art still lack sufficient understanding and solutions on how to use cloud systems and digital twin technologies to further improve the level of battery management system construction. Summary of the invention
[0003] In view of this, in order to solve the technical problems existing in the art, the present invention provides a terminal-cloud collaborative battery management method based on embedded optical fiber sensing, which specifically includes the following steps:
[0004] Step 1: For the physical entity of a specific type of single battery in the battery system to be monitored, a distributed temperature self-sensing system suitable for smart battery single cells is integrated with an embedded distributed optical fiber sensor system, and the sensor data collected by the distributed temperature self-sensing system is obtained through the terminal battery management system and uploaded to the cloud system; the cloud system stores specific cycle charge and discharge experimental data of the same type of battery single cells in the experimental environment, as well as the historical operation data of the controlled battery;
[0005] Step 2: Build an equivalent circuit model and heat collection model of the physical entity of the single battery on the terminal battery management system, and use it to collect battery current and terminal voltage data in real time; perform parameter identification on the equivalent circuit model based on the least squares method; estimate the battery state of charge SOC based on the extended Kalman filter EKF and the identified equivalent circuit model; input the SOC estimation result into the SOC-OCV relationship issued by the cloud system to obtain the open circuit voltage OCV value; use the obtained OCV value to perform parameter identification on the heat collection model; estimate the battery heat generation rate based on the extended Kalman filter EKF and the identified heat collection model;
[0006] Step 3: Establish a long short-term memory network LSTMN in the cloud system to perform online prediction of the SOC-OCV relationship corresponding to different temperature and charge-discharge rate characteristics; the training of the LSTMN specifically uses the experimental data obtained from the battery test experiment, with the battery historical temperature distribution and charge-discharge rate characteristic data as input, OCV and the corresponding SOC standard segment as output, and use polynomial fitting to obtain the final SOC-OCV relationship curve and send it to the terminal battery management system for real-time correction and update of the model parameters of the model established on the terminal battery management system;
[0007] Step 4: The cloud system uses the continuously collected test experimental data and the historical operation data of each battery cell to regularly train, adjust and update the LSTMN.
[0008] Furthermore, in step one, for the physical entity of the cylindrical battery, the integrated construction of the distributed temperature self-sensing system is achieved through the following steps in sequence: first, a hole is drilled in the center of the negative terminal of the battery cell in an inert gas environment, and its internal structure is sealed to protect it; then, the positive terminal is opened with a tube cutter and sealed and isolated to prevent short circuits, and at the same time, a distributed optical fiber is inserted from the negative terminal through the isolation layer into the battery; electrical and optical communication connections are established between the resealed battery cell and the terminal battery management system, so that all battery cells and the terminal battery management system together constitute a distributed temperature self-sensing system that can sense the internal and surface temperatures of each cell.
[0009] Furthermore, in step 2, the following equivalent circuit model including two RC circuits is specifically established for the battery cell:
[0010]
[0011] V t (t) = V oc (t)+V p1 (t)+V p2 (t)+R 0 (t)I(t)
[0012] Where I is the battery load current, R 1 and R 2 is the polarization internal resistance of the two RC loops, C 1 and C 2 is the polarized capacitance of the two RC loops, R 0 is the ohmic internal resistance, V p1 and V p2 is the voltage across the two RC loops, V OC is the open circuit voltage, V t is the terminal voltage, t is the time;
[0013] After Laplace transformation, the following discrete form is obtained:
[0014]
[0015] θ e =[k 1 k 2 k 3 k 4 k 5 ] T
[0016]
[0017] in,
[0018]
[0019] Parameter matrix θ e is the model parameter to be identified, which is obtained by the following least squares method:
[0020]
[0021] Where P(k) is the covariance matrix; K(k) is the gain matrix; e(k) is the estimation error; y is the output variable; and I is the identity matrix.
[0022] Furthermore, in step 2, a heat collection model of the following form is specifically established for the battery cell:
[0023]
[0024] Q(t)=I(t)(V oc (t)-V t (t))-λI(t)T a (t)
[0025] Among them, T c and T s are the average values of the center temperature and surface temperature of the battery cell obtained through distributed optical fiber; C cRepresents the thermal capacity of the battery cell, R c represents thermal resistance; Q is the heat generation rate, I(t) is the load current, T a It is T c and T s The average value of λI(t)T a (t) is the reversible entropy heat, where λ is the entropy coefficient;
[0026] After Laplace transformation, the following discrete form is obtained:
[0027]
[0028] θ t =[1 α β γ] T
[0029]
[0030] Among them, the parameter matrix θ t are the model parameters to be identified, P(k) is the covariance matrix; K(k) is the gain matrix; e(k) is the estimation error; y is the output variable; and I is the identity matrix.
[0031] Furthermore, in step 3, we define the neural unit in LSTMN at each time step t, and the hidden state h t From the data x at the same time step t Update, the hidden state h of the previous time step t-1 , input gate i t , input node g t , forget gate f t , output gate o t and storage unit c t , the update formula is:
[0032]
[0033] Where W and b are the layer weights and biases, respectively, W fx and W fh They are respectively the forget gate relative to x t and h t-1 The two weight matrices b f is the bias term, σ is the Sigmoid function; W ix and W ih They are respectively the input gate relative to x t and h t-1 The two weight matrices b i is the bias term; W gx and W gh They are respectively relative to x in the tanh activation function t and h t-1The two weight matrices, b g is the bias term; W ox and W oh They are the output gates relative to x. t and h t-1 The two weight matrices b o is the bias term;
[0034] A dropout layer is introduced for model training, through which some hidden outputs are randomly masked and these neurons do not affect the forward propagation during training; the estimated result of the standard open circuit voltage matrix is output based on the following formula:
[0035]
[0036] Among them, W out and b out They are the weight and bias of the discard layer, and the output value It is the OCV estimation result corresponding to the standard SOC array, and relu is selected as the activation function; for model training, the mean square error MSE is used as the model loss:
[0037]
[0038] Where n is the number of training samples;
[0039] The trained LSTMN can output the OCV corresponding to the SOC list output =[OCV 1 OCV 2 ... OCV 10 ];
[0040] The SOC-OCV relationship is obtained by fitting the following polynomial:
[0041]
[0042] Where M is the highest coefficient of the polynomial, SOC j represents the jth power of SOC, w j For SOC j The coefficient of
[0043] The fitting error is characterized by the following formula:
[0044]
[0045] Where N is the number of samples; the final polynomial coefficients are obtained by the following formula:
[0046]
[0047] Accordingly, the present invention also provides an end-cloud collaborative battery management system based on embedded optical fiber sensing for executing the above method, which is composed of a distributed temperature self-sensing system, a terminal battery management system and a cloud system;
[0048] Among them, the distributed temperature self-sensing system is used to obtain the internal and external temperature distribution of the battery physical entity;
[0049] The terminal battery management system is used to obtain the battery cell temperature data and battery operation data collected by the distributed temperature self-sensing system, perform multi-dimensional estimation of the battery SOC and heat generation rate through the physical entity equivalent circuit model and heat collection model built on it, and upload the collected raw data and the estimated battery status results to the cloud system;
[0050] The cloud system is used to store and process the data uploaded by the terminal battery management, and use the battery's historical temperature distribution and charge and discharge rate characteristic data and test experiment data to train the LSTMN network, and apply the trained LSTMN network online to estimate the SOC-OCV relationship; and send the obtained SOC-OCV relationship to the terminal battery management system for correcting and updating the model parameters.
[0051] The end-cloud collaborative battery management method and system based on embedded optical fiber sensing provided by the present invention integrates a distributed temperature self-sensing system at the battery end to sense the temperature distribution inside and outside the battery cell, performs battery modeling and state estimation in the battery management system, and estimates the SOC-OCV curve in the cloud to guide the correction and update of the terminal model, thereby realizing the multi-level organic integration of smart battery end, terminal and cloud data, making the model have higher accuracy and wider adaptability, and can also ensure the richness of state estimation items, which significantly improves the accuracy and efficiency of battery electrical-thermal management compared with the existing technology. BRIEF DESCRIPTION OF THE DRAWINGS
[0052] Figure 1 An optional structural block diagram of the system provided by the present invention;
[0053] Figure 2 This is a flow chart of model parameter identification state estimation of the present invention.
[0054] Figure 3 A flow chart for integrating and constructing the distributed temperature self-sensing system of the present invention;
[0055] Figure 4 The following is an optional flow chart for conducting a test experiment on a battery according to an example of the present invention. DETAILED DESCRIPTION
[0056] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0057] The end-cloud collaborative battery management method based on embedded optical fiber sensing provided by the present invention is as follows: Figure 1 , 2 As shown, the specific steps include:
[0058] Step 1: For the physical entity of a specific type of single battery in the battery system to be monitored, a distributed temperature self-sensing system suitable for smart battery single cells is integrated with an embedded distributed optical fiber sensor system, and the sensor data collected by the distributed temperature self-sensing system is obtained through the terminal battery management system and uploaded to the cloud system; the cloud system stores specific cycle charge and discharge experimental data of the same type of battery single cells in the experimental environment, as well as the historical operation data of the controlled battery;
[0059] Step 2: Build an equivalent circuit model and heat collection model of the physical entity of the single battery on the terminal battery management system, and use it to collect battery current and terminal voltage data in real time; perform parameter identification on the equivalent circuit model based on the least squares method; estimate the battery state of charge SOC based on the extended Kalman filter EKF and the identified equivalent circuit model; input the SOC estimation result into the SOC-OCV relationship issued by the cloud system to obtain the open circuit voltage OCV value; use the obtained OCV value to perform parameter identification on the heat collection model; estimate the battery heat generation rate based on the extended Kalman filter EKF and the identified heat collection model;
[0060] Step 3: Establish a long short-term memory network LSTMN in the cloud system to perform online prediction of the SOC-OCV relationship corresponding to different temperature and charge-discharge rate characteristics; the training of the LSTMN is specifically performed by Figure 4 The experimental data obtained from the battery test experiment shown in the figure takes the battery historical temperature distribution and charge and discharge rate characteristic data as input, and OCV and the corresponding SOC standard segment as output. The final SOC-OCV relationship curve is obtained by polynomial fitting and sent to the terminal battery management system for real-time correction and update of the model parameters of the model established on the terminal battery management system;
[0061] Step 4: The cloud system uses the continuously collected test experimental data and the historical operation data of each battery cell to regularly train, adjust and update the LSTMN.
[0062] In a preferred embodiment of the present invention, in step 1, for the physical entity of the cylindrical battery, the following steps are performed in sequence: Figure 3 The steps shown realize the integrated construction of the distributed temperature self-sensing system: first, a hole with a diameter of 0.9 mm is drilled in the center of the negative terminal of the battery cell in an inert gas environment, and its internal structure is sealed with Kapton tape to protect it; then the positive terminal is opened with a tube cutter and sealed with Kapton tape to prevent short circuit, and at the same time, the distributed optical fiber is inserted from the negative terminal through the isolation layer into the battery; the resealed battery cell is connected to the terminal battery management system through electrical and optical communication, so that all battery cells and the terminal battery management system together constitute a distributed temperature self-sensing system that can sense the internal and surface temperature of each cell.
[0063] In a preferred embodiment of the present invention, in step 2, the following equivalent circuit model including two RC circuits is specifically established for the battery cell:
[0064]
[0065] V t (t) = V oc (t)+V p1 (t)+V p2 (t)+R 0 (t)I(t)
[0066] Where I is the battery load current, R 1 and R 2 is the polarization internal resistance of the two RC loops, C 1 and C 2 is the polarized capacitance of the two RC loops, R 0 is the ohmic internal resistance, V p1 and V p2 is the voltage across the two RC loops, V OC is the open circuit voltage, V t is the terminal voltage, t is the time;
[0067] After Laplace transformation, the following discrete form is obtained:
[0068]
[0069] θ e =[k 1 k 2 k 3 k 4 k 5 ] T
[0070]
[0071] in,
[0072]
[0073] Parameter matrix θ e is the model parameter to be identified, which is obtained by the following least squares method:
[0074]
[0075] Where P(k) is the covariance matrix; K(k) is the gain matrix; e(k) is the estimation error; y is the output variable; and I is the identity matrix.
[0076] In a preferred embodiment of the present invention, in step 2, a heat collection model of the following form is specifically established for the battery cell:
[0077]
[0078] Q(t)=I(t)(V oc (t)-V t (t))-λI(t)T a (t)
[0079] Among them, T c and T s are the average values of the center temperature and surface temperature of the battery cell obtained through distributed optical fiber; C c Represents the thermal capacity of the battery cell, R c represents thermal resistance; Q is the heat generation rate, I(t) is the load current, T a It is T c and T s The average value of λI(t)T a (t) is the reversible entropy heat, where λ is the entropy coefficient;
[0080] After Laplace transformation, the following discrete form is obtained:
[0081]
[0082] θ t =[1αβγ] T
[0083]
[0084] Among them, the parameter matrix θ t are the model parameters to be identified, P(k) is the covariance matrix; K(k) is the gain matrix; e(k) is the estimation error; y is the output variable; and I is the identity matrix.
[0085] The equivalent circuit model can be used to perform the following extended Kalman filter EKF algorithm to estimate the battery SOC"
[0086]
[0087] In a preferred embodiment of the present invention, step 3 specifically defines that at each time step t, the hidden state h of the neural unit in the LSTMN is t From the data x at the same time step t Update, the hidden state h of the previous time step t-1 , input gate i t , input node g t , forget gate f t , output gate o t and storage unit c t , the update formula is:
[0088]
[0089] Where W and b are the layer weights and biases, respectively, W fx and W fh They are respectively the forget gate relative to x t and h t-1 The two weight matrices b f is the bias term, σ is the Sigmoid function; W ix and W ih They are respectively the input gate relative to x t and h t-1 The two weight matrices b i is the bias term; W gx and W gh They are respectively relative to x in the tanh activation function t and h t-1 The two weight matrices b g is the bias term; W ox and W oh They are the output gates relative to x. t and h t-1 The two weight matrices b o is the bias term;
[0090] A dropout layer is introduced for model training, through which some hidden outputs are randomly masked and these neurons do not affect the forward propagation during training; the estimated result of the standard open circuit voltage matrix is output based on the following formula:
[0091]
[0092] Among them, W out and b out are the discard layer weights and biases, and the output values It is the OCV estimation result corresponding to the standard SOC array, and relu is selected as the activation function; for model training, the mean square error MSE is used as the model loss:
[0093]
[0094] Where n is the number of training samples;
[0095] The trained LSTMN can output the OCV corresponding to the SOC list output =[OCV 1 OCV 2 ... OCV 10 ];
[0096] The SOC-OCV relationship is obtained by fitting the following polynomial:
[0097]
[0098] Wherein, M is the highest coefficient of the polynomial. In this embodiment, M=6 is set, SOC j represents the jth power of SOC, w j For SOC j The coefficient of
[0099] The fitting error is characterized by the following formula:
[0100]
[0101] Where N is the number of samples, and in this embodiment, N=10 is set; the final polynomial coefficients are obtained by the following formula:
[0102]
[0103] Accordingly, the present invention also provides an end-cloud collaborative battery management system based on embedded optical fiber sensing for executing the above method, which is composed of a distributed temperature self-sensing system, a terminal battery management system and a cloud system;
[0104] Among them, the distributed temperature self-sensing system is used to obtain the internal and external temperature distribution of the battery physical entity;
[0105] The terminal battery management system is used to obtain the battery cell temperature data and battery operation data collected by the distributed temperature self-sensing system, perform multi-dimensional estimation of the battery SOC and heat generation rate through the physical entity equivalent circuit model and heat collection model built on it, and upload the collected raw data and the estimated battery status results to the cloud system;
[0106] The cloud system is used to store and process the data uploaded by the terminal battery management, and use the battery's historical temperature distribution and charge and discharge rate characteristic data and test experiment data to train the LSTMN network, and apply the trained LSTMN network online to estimate the SOC-OCV relationship; and send the obtained SOC-OCV relationship to the terminal battery management system for correcting and updating the model parameters.
[0107] In specific implementation, the terminal battery management system can be manufactured using, but not limited to, hardware unit modules, display modules, communication modules, and related wiring harnesses and structural parts (covers, brackets, bolts, etc.). Its operating model and corresponding SOC and battery heat generation rate estimation algorithms can be developed and implemented based on the battery management system developed by the charger or embedded system. The cloud system may include, but is not limited to, data transmission modules, cloud computing modules, data storage modules, etc. The distributed temperature self-sensing system and the terminal battery management system can exchange data through methods including but not limited to the control local area network, and the terminal battery management system and the cloud system can exchange data through wireless communication technologies including but not limited to ZigBee, WIFI, 3G / 4G / 5G, etc.
[0108] It should be understood that the size of the serial numbers of the steps in the embodiment of the present invention does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention.
[0109] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.
Claims
1. The end-cloud collaborative battery management method based on embedded optical fiber sensing is characterized by: The specific steps include: Step 1: For the physical entity of a specific type of single battery in the battery system to be monitored, a distributed temperature self-sensing system suitable for smart battery single cells is integrated with an embedded distributed optical fiber sensor system, and the sensor data collected by the distributed temperature self-sensing system is obtained through the terminal battery management system and uploaded to the cloud system; The cloud system stores specific cycle charge and discharge experimental data of the same type of battery cells in the experimental environment, as well as the historical operation data of the controlled battery; Step 2: Build an equivalent circuit model and heat collection model of the physical entity of the single battery on the terminal battery management system, and use it to collect battery current and terminal voltage data in real time; perform parameter identification on the equivalent circuit model based on the least squares method; The battery state of charge (SOC) is estimated based on the extended Kalman filter (EKF) and the identified equivalent circuit model; the SOC estimation result is input into the SOC-OCV relationship issued by the cloud system to obtain the open circuit voltage (OCV) value; the obtained OCV value is used to identify the parameters of the thermal collector model; The heat generation rate of the battery is estimated based on the extended Kalman filter (EKF) and the identified heat collection model; Step 3: Establish a long short-term memory network LSTMN in the cloud system to perform online prediction of the SOC-OCV relationship corresponding to different temperature and charge-discharge rate characteristics; the training of the LSTMN specifically uses the experimental data obtained from the battery test experiment, with the battery historical temperature distribution and charge-discharge rate characteristic data as input, OCV and the corresponding SOC standard segment as output, and use polynomial fitting to obtain the final SOC-OCV relationship curve and send it to the terminal battery management system for real-time correction and update of the model parameters of the model established on the terminal battery management system; Step 4: The cloud system uses the continuously collected test experimental data and the historical operation data of each battery cell to regularly train, adjust and update the LSTMN.
2. The method according to claim 1, characterized in that: In step one, for the physical entity of the cylindrical battery, the integrated construction of the distributed temperature self-sensing system is achieved through the following steps in sequence: first, a hole is drilled in the center of the negative terminal of the battery cell in an inert gas environment, and its internal structure is sealed to protect it; then, the positive terminal is opened with a tube cutter and sealed and isolated to prevent short circuits, and at the same time, the distributed optical fiber is inserted from the negative terminal through the isolation layer into the battery; the resealed battery cell is connected to the terminal battery management system through electrical and optical communications, so that all battery cells and the terminal battery management system together constitute a distributed temperature self-sensing system that can sense the internal and surface temperatures of each cell.
3. The method according to claim 1, characterized in that: In step 2, the following equivalent circuit model including two RC circuits is specifically established for the battery cell: V t (t)=V oc (t)+V p1 (t)+V p2 (t)+R0(t)I(t) Where I is the battery load current, R1 and R2 are the polarization internal resistances of the two RC loops, C1 and C2 are the polarization capacitances of the two RC loops, R0 is the ohmic internal resistance, and V p1 and V p2 is the voltage across the two RC loops, V OC is the open circuit voltage, V t is the terminal voltage, t is the time; After Laplace transformation, the following discrete form is obtained: θ e =[k1 k2 k3 k4 k5] T in, Parameter matrix θ e is the model parameter to be identified, which is obtained by the following least squares method: Where P(k) is the covariance matrix; K(k) is the gain matrix; e(k) is the estimation error; y is the output variable; and I is the identity matrix.
4. The method according to claim 1, characterized in that: In step 2, the following heat collection model is established for the battery cell: Q(t)=I(t)(V oc (t)-V t (t))-λI(t)T a (t) Among them, T c and T s are the average values of the center temperature and surface temperature of the battery cell obtained through distributed optical fiber; C c Represents the thermal capacity of the battery cell, R c represents thermal resistance; Q is the heat generation rate, I(t) is the load current, T a It is T c and T s The average value of λI(t)T a (t) is the reversible entropy heat, where λ is the entropy coefficient; After Laplace transformation, the following discrete form is obtained: i t =[1a bg] T Among them, the parameter matrix θ t are the model parameters to be identified, P(k) is the covariance matrix; K(k) is the gain matrix; e(k) is the estimation error; y is the output variable; and I is the identity matrix.
5. The method according to claim 1, characterized in that: In step 3, we specifically define the neural unit in LSTMN at each time step t, and the hidden state h t From the data x at the same time step t Update, the hidden state h of the previous time step t-1 , input gate i t , input node g t , forget gate f t , output gate o t and storage unit c t , the update formula is: Where W and b are the layer weights and biases, respectively, W fx and W fh They are respectively the forget gate relative to x t and h t-1 The two weight matrices b f is the bias term, σ is the Sigmoid function; W ix and W ih They are respectively the input gate relative to x t and h t-1 The two weight matrices b i is the bias term; W gx and W gh They are respectively the tanh activation function relative to x t and h t-1 The two weight matrices b g is the bias term; W ox and W oh They are the output gates relative to x. t and h t-1 The two weight matrices b o is the bias term; A dropout layer is introduced for model training, through which some hidden outputs are randomly masked and these neurons do not affect the forward propagation during training; Output the estimated result of the standard open circuit voltage matrix based on the following formula: Among them, W out and b out are the discard layer weights and biases, and the output values It is the OCV estimation result corresponding to the standard SOC array, and relu is selected as the activation function; for model training, the mean square error MSE is used as the model loss: Where n is the number of training samples; The trained LSTMN can output the OCV corresponding to the SOC list output =[OCV1 OCV2...OCV 10 ]; The SOC-OCV relationship is obtained by fitting the following polynomial: Where M is the highest coefficient of the polynomial, SOC j represents the jth power of SOC, w j For SOC j The coefficient of The fitting error is characterized by the following formula: Where N is the number of samples; the final polynomial coefficients are obtained by the following formula:
6. A terminal-cloud collaborative battery management system based on embedded optical fiber sensing for executing the method according to any one of claims 1 to 5, characterized in that: It consists of a distributed temperature self-sensing system, a terminal battery management system, and a cloud system; Among them, the distributed temperature self-sensing system is used to obtain the internal and external temperature distribution of the battery physical entity; The terminal battery management system is used to obtain the battery cell temperature data and battery operation data collected by the distributed temperature self-sensing system, perform multi-dimensional estimation of the battery SOC and heat generation rate through the physical entity equivalent circuit model and heat collection model built on it, and upload the collected raw data and the estimated battery status results to the cloud system; The cloud system is used to store and process the data uploaded by the terminal battery management, and use the battery's historical temperature distribution and charge and discharge rate characteristic data and test experiment data to train the LSTMN network, and apply the trained LSTMN network online to estimate the SOC-OCV relationship; and send the obtained SOC-OCV relationship to the terminal battery management system for correcting and updating the model parameters.
Citation Information
Patent Citations
Energy storage device operation situation deduction method based on LSTM
CN113049962A
Battery surface temperature distribution estimation method and system
CN113884901A
Power battery fault monitoring method and system based on information physical fusion technology
CN113960476A
One-dimensional joint simulation modeling method for thermal management system of hybrid electric vehicle
CN114239133A
Advanced Ground Thermal Conductivity Testing
US20220196583A1
Cited By
Joint estimation method for external thermal resistance and internal and external temperatures of lithium ion power battery
CN119438913A
Walking control method and device of battery sampling line breakage fault-tolerant control system and electronic equipment
CN120645696A
