Battery life prediction method, device and system based on smart metering electric energy meter

Abstract

The application discloses a kind of based on intelligent interconnection electric energy meter battery life prediction method, device and system, method includes: through intelligent interconnection electric energy meter, the charging data of battery to be measured in charging time period is obtained;In the case where the charging data is not matched with local model in intelligent interconnection electric energy meter, charging data is transmitted to main station system;Main station system is based on the capacity specification of battery to be measured, corresponding weight is configured to charging voltage, current steady-state quantity, current transient quantity and voltage current waveform respectively;Based on the weight allocated, the weight parameter of trained neural network model is adjusted correspondingly;The difference of charging voltage, current steady-state quantity, current transient quantity and voltage current waveform and corresponding standard charging voltage, standard current steady-state quantity, standard current transient quantity and standard voltage current waveform is calculated respectively;Difference is as the input of neural network model, and the life data of battery to be measured is obtained by output, and then the life of the battery can be known.

Description

Technical Field

[0001] This invention relates to the field of electricity meters and electricity consumption information collection technology, and in particular to a method, device and system for predicting battery life based on smart IoT electricity meters. Background Technology

[0002] Under normal circumstances, electric vehicle batteries have a long lifespan. However, with the increasing prevalence of electric vehicles and their growing overall number, the question of how to predict battery life and safety arises. Recently, electric vehicle fires have become increasingly common, generally caused by battery failure. To avoid this problem, current methods typically include:

[0003] 1. The car manufacturer has notified the customer to replace the battery when the car reaches the end of its manufacturing life.

[0004] 2. During maintenance, check the battery status and replace the battery if any abnormalities are found.

[0005] The problems mentioned above are:

[0006] 1. Due to factors such as car owner usage habits and battery quality control, many car batteries may reach the end of their lifespan before reaching their designated lifespan.

[0007] 2. Car owners may not perform maintenance according to the standard schedule, so potential battery hazards still exist.

[0008] 3. Even if the car owner takes the car for maintenance and performs the corresponding battery test, it can only detect batteries that already have problems, but cannot predict the battery's lifespan.

[0009] Therefore, there is a need for a way to predict battery life during the normal operation of electric vehicles, especially during charging. Summary of the Invention

[0010] This invention provides a method, device, and system for predicting battery life based on a smart IoT energy meter, which has the function of detecting battery life during charging.

[0011] This invention provides a battery life prediction method based on a smart IoT energy meter, the method comprising:

[0012] The charging data of the battery under test during the charging period is obtained by a smart IoT energy meter. The charging data includes charging voltage, steady-state current, transient current, and voltage and current waveforms.

[0013] The smart IoT energy meter performs data matching between the charging data and the local model; if the matching fails, the charging data is transmitted to the main station system.

[0014] The master station system assigns corresponding weights to the charging voltage, steady-state current, transient current, and voltage and current waveforms based on the capacity specifications of the battery under test.

[0015] Based on the assigned weights, the weight parameters of the trained neural network model are adjusted accordingly.

[0016] The standard charging voltage, standard steady-state current, standard transient current, and standard voltage and current waveforms of the battery under test are obtained in advance.

[0017] Calculate the differences between the charging voltage, steady-state current, transient current, and voltage and current waveforms and the corresponding standard charging voltage, standard steady-state current, standard transient current, and standard voltage and current waveforms.

[0018] The difference is used as the input to the neural network model, and the output is the lifespan data of the battery under test.

[0019] In one possible implementation, the step of assigning corresponding weights to the charging voltage, steady-state current, transient current, and voltage-current waveforms based on the capacity specifications of the battery under test includes:

[0020] If the capacity specification of the battery under test is greater than or equal to a preset threshold T, then the weight of the charging voltage is set to... The weight of the steady-state current is set as follows: The weight of the current transient is set as follows: The weights of the voltage and current waveforms are set as follows: Among them, the preset threshold T is greater than 1;

[0021] If the battery capacity is less than a preset threshold T, then the weight of the charging voltage is set to... The weight of the steady-state current is set as follows: The weight of the current transient is set as follows: The weights of the voltage and current waveforms are set as follows:

[0022] In one possible implementation, the preset threshold T is set to 1.5.

[0023] In one possible implementation, adjusting the weight parameters of the trained neural network model based on the assigned weights includes:

[0024] Find the smallest weight parameter Qn among all weight parameters in the neural network model other than the weight parameters of charging voltage, steady-state current, transient current, and voltage and current waveforms.

[0025] The weight parameter Q1 of the charging voltage in the neural network model is x1*(1+0.25Qn);

[0026] The weight parameter Q2 = x2*(1+0.25Qn) for the steady-state current in the neural network model;

[0027] The weight parameter Q3 = x3*(1+0.25Qn) for the transient current in the neural network model;

[0028] The weighting parameter Q4 for the voltage and current waveforms in the neural network model is x4*(1+0.25Qn);

[0029] Remove the weight parameter Qn from the neural network model.

[0030] In one possible implementation, after outputting the lifespan data of the battery under test, the method further includes:

[0031] The neural network model in the master station system is sent to the smart IoT energy meter.

[0032] In one possible implementation, after outputting the lifespan data of the battery under test, the method further includes:

[0033] If the capacity specification of the battery under test is between 15 and 30 kWh, then the abnormal life threshold of the battery under test is set to 50%.

[0034] If the capacity specification of the battery under test is between 30 and 60 kWh, then the abnormal life threshold of the battery under test is set to 60%.

[0035] If the lifespan data is lower than the lifespan abnormality threshold, the battery under test is determined to be abnormal.

[0036] If the battery under test is determined to be abnormal, the corresponding abnormality information will be notified to the vehicle owner.

[0037] Another aspect of the present invention provides a battery life prediction system based on a smart IoT energy meter, the system comprising an electric vehicle charging pile, a smart IoT energy meter, an energy controller and a master station system connected in sequence;

[0038] During the charging process of the electric vehicle charging pile and the electric vehicle, the smart IoT energy meter acquires the charging data of the battery under test during the charging time period. The charging data includes charging voltage, steady-state current, transient current, and voltage and current waveforms.

[0039] The smart IoT energy meter performs data matching between the charging data and the local model. If the matching fails, the charging data is transmitted to the main station system through the energy controller.

[0040] The main station system assigns corresponding weights to the charging voltage, steady-state current, transient current, and battery temperature-voltage-current waveforms based on the capacity specifications of the battery under test. Based on the assigned weights, it adjusts the weight parameters of the trained neural network model accordingly. It pre-acquires the standard charging voltage, standard steady-state current, standard transient current, and standard battery temperature-voltage-current waveforms of the battery under test. It calculates the differences between the charging voltage, steady-state current, transient current, and battery temperature-voltage-current waveforms and their corresponding standard values. These differences are used as input to the neural network model, and the system outputs the lifespan data of the battery under test.

[0041] In one implementation, the master station system will also send the neural network model with adjusted weight parameters to the smart IoT energy meter through the energy controller.

[0042] In one embodiment, the smart IoT energy meter includes a voltage and current waveform acquisition module, a management and control unit, and a data transmission module that are connected in sequence.

[0043] During the charging process of the electric vehicle charging pile and the electric vehicle, the voltage and current waveform acquisition module acquires the voltage and current waveform of the battery under test during the charging period and transmits the voltage and current waveform to the data transmission module through the management and control unit;

[0044] The data transmission module transmits the voltage and current waveforms to the master station system.

[0045] Another aspect of the present invention provides a battery life prediction device based on a smart IoT energy meter, the device comprising:

[0046] The first acquisition module is used to acquire the charging data of the battery under test during the charging period through a smart IoT energy meter. The charging data includes charging voltage, steady-state current, transient current, and voltage and current waveforms.

[0047] The local analysis module is used to perform data matching between the charging data and the local model based on the charging data in the smart IoT energy meter; if the matching fails, the charging data is transmitted to the main station system.

[0048] The allocation module is used to enable the master station system to configure corresponding weights for the charging voltage, steady-state current, transient current, and voltage and current waveforms based on the capacity specifications of the battery under test.

[0049] The adjustment module is used to adjust the weight parameters of the trained neural network model based on the assigned weights.

[0050] The second acquisition module is used to acquire in advance the standard charging voltage, standard current steady-state quantity, standard current transient quantity, and standard voltage and current waveform of the battery under test;

[0051] The calculation module is used to calculate the differences between the charging voltage, steady-state current, transient current, and voltage and current waveforms and the corresponding standard charging voltage, standard steady-state current, standard transient current, and standard voltage and current waveforms, respectively.

[0052] The prediction module is used to take the difference as input to the neural network model and output the life data of the battery under test.

[0053] In this embodiment of the invention, the application acquires charging data of the battery under test during the charging period; based on the capacity specifications of the battery under test, it assigns corresponding weights to the charging voltage, steady-state current, transient current, and voltage-current waveform; based on the assigned weights, it adjusts the weight parameters of the trained neural network model accordingly; it pre-acquires the standard charging voltage, standard steady-state current, standard transient current, and standard voltage-current waveform of the battery under test; it calculates the differences between the charging voltage, steady-state current, transient current, and voltage-current waveform and the corresponding standard charging voltage, standard steady-state current, standard transient current, and standard voltage-current waveform; and it uses the differences as input to the neural network model to output the lifespan data of the battery under test. This solution can predict the battery's lifespan based on the charging voltage, steady-state current, transient current, and voltage-current waveform during charging, allowing vehicle owners to understand the vehicle's lifespan each time it is charged and take timely action. Attached Figure Description

[0054] The above and other objects, features, and advantages of exemplary embodiments of the present invention will become readily apparent from the following detailed description taken in conjunction with the accompanying drawings. Several embodiments of the invention are illustrated in the drawings by way of example and not limitation, wherein:

[0055] In the accompanying drawings, the same or corresponding reference numerals indicate the same or corresponding parts.

[0056] Figure 1 This is a schematic diagram illustrating the implementation process of a battery life prediction method based on a smart IoT energy meter according to an embodiment of the present invention.

[0057] Figure 2 This is a network topology diagram of a battery life prediction system based on a smart IoT energy meter, according to an embodiment of the present invention.

[0058] Figure 3 This is a structural diagram of a battery life prediction device based on a smart IoT energy meter, according to an embodiment of the present invention. Detailed Implementation

[0059] To make the objectives, features, and advantages of this invention more apparent and understandable, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0060] like Figure 1 As shown, this invention provides a method for predicting battery life based on a smart IoT energy meter, the method comprising:

[0061] Step 101: Obtain the charging data of the battery under test during the charging period through the smart IoT energy meter. The charging data includes charging voltage, steady-state current, transient current, and voltage and current waveforms.

[0062] Step 102: Match the charging data with the local model in the smart IoT energy meter; if the matching fails, transmit the charging data to the main station system.

[0063] Step 103: Based on the capacity specifications of the battery under test, the main station system configures corresponding weights for the charging voltage, steady-state current, transient current, and voltage and current waveforms.

[0064] Step 104: Based on the assigned weights, adjust the weight parameters of the trained neural network model accordingly.

[0065] Step 105: Pre-acquire the standard charging voltage, standard steady-state current, standard transient current, and standard voltage and current waveforms of the battery under test;

[0066] Step 106: Calculate the differences between the charging voltage, steady-state current, transient current, and voltage and current waveforms and the corresponding standard charging voltage, standard steady-state current, standard transient current, and standard voltage and current waveforms.

[0067] Step 107: Use the difference as input to the neural network model and output the life data of the battery under test.

[0068] In this embodiment, the solution can be applied to a charging pile system, wherein, reference Figure 2 As shown, the charging pile system includes an electric vehicle charging pile, a smart IoT energy meter, an energy controller, and a master station system that are connected in sequence.

[0069] Charging piles are used to connect to electric vehicles and charge them; smart IoT energy meters are used to record charging data, including charging voltage, steady-state current, and transient current.

[0070] The smart IoT energy meter further includes a voltage and current waveform acquisition module, a management control unit, and a data transmission module that are connected in sequence. During the charging process between the charging pile and the electric vehicle, the voltage and current waveform acquisition module acquires the voltage and current waveforms of the battery under test during the charging period and transmits the voltage and current waveforms to the data transmission module through the management control unit. The data transmission module transmits the voltage and current waveforms to the main station system.

[0071] Therefore, in step 101, the charging data can be obtained from the IoT energy meter, and the IoT energy meter transmits the charging data to the energy controller and the main station system in sequence.

[0072] In step 102, after acquiring the charging data, local data analysis is first performed on the smart IoT energy meter. The smart IoT energy meter pre-stores some data models related to the charging data, such as pre-stored multiple sets of data. Each set of data includes charging voltage, steady-state current, transient current, voltage and current waveforms, and corresponding battery life data. If the acquired charging data matches the data model, the battery life data can be directly obtained. If they do not match, the charging data is transmitted to the main station system through the energy controller, and the main station system performs further data analysis.

[0073] In step 103, since the charging voltage, steady-state current, transient current, and voltage-current waveforms all affect battery life, a neural network model is needed to predict battery life based on these four factors. According to their importance to the battery, the charging voltage, steady-state current, transient current, and voltage-current waveforms need to be weighted, so that factors with higher influence are assigned higher weights.

[0074] In step 104, the neural network model includes: an input layer, a convolutional layer, a fusion layer, a fully connected layer, and an output layer.

[0075] The weight parameters are located in the fully connected layer. Based on the pre-assigned weights of the charging voltage, steady-state current, transient current, and voltage and current waveforms, the weight parameters of the fully connected layer are adjusted accordingly to make the output results more accurate.

[0076] In step 105, the standard charging voltage, standard steady-state current, standard transient current, and standard voltage and current waveforms of the battery under test are the data when the battery is working normally or just leaving the factory. These data can be obtained by consulting relevant materials.

[0077] In step 106, the current charging voltage of the battery under test is subtracted from the standard charging voltage to obtain the voltage difference. Similarly, the steady-state current difference, transient current difference, and temperature difference can also be calculated. The purpose of this step is to enable the subsequent neural network model to integrate the current data and standard data for prediction, resulting in more accurate and universal output results.

[0078] In step 107, the difference is used as the input to the neural network model, and the output is the life data of the battery under test.

[0079] Therefore, by acquiring charging data of the battery under test during the charging period; when the charging data fails to match the local model data, the main station system assigns corresponding weights to the charging voltage, steady-state current, transient current, and voltage-current waveforms based on the capacity specifications of the battery under test; based on the assigned weights, the weight parameters of the trained neural network model are adjusted accordingly; the standard charging voltage, standard steady-state current, standard transient current, and standard voltage-current waveforms of the battery under test are pre-acquired; the differences between the charging voltage, steady-state current, transient current, and voltage-current waveforms and the corresponding standard charging voltage, standard steady-state current, standard transient current, and standard voltage-current waveforms are calculated; the differences are used as input to the neural network model, and the output yields the battery life data. This solution can predict the battery life based on the charging voltage, steady-state current, transient current, and voltage-current waveforms during charging, allowing vehicle owners to understand the vehicle's lifespan each time it is charged and take timely action.

[0080] In one possible implementation, based on the capacity specifications of the battery under test, corresponding weights are assigned to the charging voltage, steady-state current, transient current, and voltage-current waveforms, including:

[0081] If the capacity specification of the battery under test is greater than the preset threshold T, then the weight of the charging voltage is set to... The weights of the steady-state current quantities are set as follows: The weights of the transient current quantities are set as follows: The weights of the voltage and current waveforms are set to... Among them, the preset threshold T is greater than 1;

[0082] If the battery capacity is less than a preset threshold T, the weight of the charging voltage is set to... The weights of the steady-state current quantities are set as follows: The weights of the transient current quantities are set as follows: The weights of the voltage and current waveforms are set to...

[0083] In this embodiment, the weight allocation method in step 102 is as follows:

[0084] First, obtain the capacity specifications of the battery to be tested, which can be obtained from the battery's model specifications.

[0085] Next, the capacity specification and the preset threshold T are compared. If the capacity specification is greater than or equal to the preset threshold T, generally, the parameters affecting the battery from largest to smallest are charging voltage, steady-state current, transient current, and voltage-current waveforms. Therefore, the weighting method is to set the weight of charging voltage as follows: The weights of the steady-state current quantities are set as follows: The weights of the transient current quantities are set as follows: The weights of the voltage and current waveforms are set to... Conversely, if the capacity specification is less than the preset threshold T, generally, the parameters affecting the battery from largest to smallest are voltage and current waveforms, transient current, steady-state current, and charging voltage. Therefore, the weighting method is to set the weight of charging voltage as follows: The weights of the steady-state current quantities are set as follows: The weights of the transient current quantities are set as follows: The weights of the voltage and current waveforms are set to... Preferably, the preset threshold T is set to 1.5.

[0086] Through experiments, by setting the above weighting parameters, the predicted lifespan basically matches the actual lifespan.

[0087] In one possible implementation, the weight parameters of the trained neural network model are adjusted accordingly based on the assigned weights, including:

[0088] Find the smallest weight parameter Qn among all weight parameters in the neural network model other than the weight parameters of charging voltage, steady-state current, transient current, and voltage and current waveforms.

[0089] In the neural network model, the weight parameter for the charging voltage is Q1 = x1 * (1 + 0.25Qn);

[0090] The weighting parameter for the steady-state current in the neural network model is Q2 = x2*(1+0.25Qn);

[0091] The weighting parameter for the transient current in the neural network model is Q3 = x3*(1+0.25Qn);

[0092] The weighting parameter for the voltage and current waveforms in the neural network model is Q4 = x4*(1+0.25Qn);

[0093] Remove the weight parameter Qn from the neural network model.

[0094] In this embodiment, after the above weight allocation, the weight parameters of the trained neural network model are adjusted accordingly. The specific adjustment process is as follows:

[0095] From the fully connected layers of the neural network model, find the smallest weight parameter Qn among all weight parameters other than charging voltage, steady-state current, transient current, and voltage and current waveform weight parameters; delete the smallest weight parameter and proportionally distribute it to the charging voltage, steady-state current, transient current, and voltage and current waveform parameters that have a greater impact on battery life.

[0096] The specific allocation method is a proportional allocation, that is, adding a weight of 0.25Qn to the charging voltage, steady-state current, transient current, and voltage and current waveform parameters.

[0097] By eliminating the weight parameters with the least impact and increasing the weight parameters for charging voltage, steady-state current, transient current, and voltage and current waveforms, the results predicted by the neural network model are made more accurate.

[0098] In one possible implementation, after outputting the lifespan data of the battery under test, the method further includes:

[0099] Based on lifespan data, determine whether the battery under test is abnormal;

[0100] If the battery under test is determined to be abnormal, the corresponding abnormality information will be notified to the vehicle owner.

[0101] In this embodiment, after outputting the lifespan data of the battery under test, it is determined whether the battery under test is abnormal.

[0102] If an abnormality is detected, the corresponding abnormality information should be notified to the car owner via SMS, mini-program, or telephone to serve as a reminder.

[0103] In one possible implementation, determining whether the battery under test is abnormal based on lifetime data includes:

[0104] Based on the capacity specifications of the battery under test, determine the critical value for abnormal lifespan of the battery under test in advance;

[0105] If the lifespan data is below the abnormal lifespan threshold, the battery under test is determined to be abnormal.

[0106] In this embodiment, the specific process for determining whether the battery under test is abnormal is as follows:

[0107] Before testing, check the capacity specifications of the battery to be tested. If the capacity specifications of the battery to be tested are between 15 and 30 kWh, then the abnormal lifespan threshold of the battery to be tested is set to 50%.

[0108] If the capacity specification of the battery under test is between 30 and 60 kWh, the abnormal life threshold of the battery under test is set to 60%.

[0109] Next, determine whether the lifespan data is below the abnormal lifespan threshold. If so, the battery under test is determined to be abnormal.

[0110] In one possible implementation, after outputting the lifespan data of the battery under test, the method further includes:

[0111] Record the number of times the electric vehicle is charged and upload the electric vehicle model data to the cloud.

[0112] In this embodiment, after obtaining the battery life data, the number of times the electric vehicle is charged is recorded, and this data, along with the electric vehicle's model information, is uploaded to the cloud. Vehicle owners can then access the cloud to track the number of times their vehicle is charged, allowing them to estimate the vehicle's lifespan.

[0113] In one possible implementation, after outputting the lifespan data of the battery under test, the method further includes:

[0114] The neural network model in the master station system is sent to the smart IoT energy meter.

[0115] In this embodiment, by sending the neural network model to the smart IoT energy meter, the battery life can be predicted directly in the smart IoT energy meter after the charging data is obtained, without the need for data analysis through the main station system, thus saving the system resources of the main station system.

[0116] Another aspect of this application provides a battery life prediction system based on a smart IoT energy meter. The system includes an electric vehicle charging pile, a smart IoT energy meter, an energy controller, and a master station system that are connected in sequence.

[0117] During the charging process of electric vehicle charging stations and electric vehicles, smart IoT energy meters acquire charging data of the battery under test during the charging period. The charging data includes charging voltage, steady-state current, transient current, and voltage and current waveforms.

[0118] The smart IoT energy meter matches charging data with the local model. If the matching fails, the charging data is transmitted to the main station system through the energy controller.

[0119] Based on the capacity specifications of the battery under test, the main station system assigns corresponding weights to the charging voltage, steady-state current, transient current, and battery temperature-voltage-current waveforms. Based on the assigned weights, it adjusts the weight parameters of the trained neural network model accordingly. It pre-acquires the standard charging voltage, standard steady-state current, standard transient current, and standard battery temperature-voltage-current waveforms of the battery under test. It then calculates the differences between the charging voltage, steady-state current, transient current, and battery temperature-voltage-current waveforms and their corresponding standard values. These differences are used as input to the neural network model, and the system outputs the battery's lifespan data.

[0120] In this embodiment, charging data of the battery under test during the charging period is obtained through a smart IoT energy meter. If the charging data fails to match the local model data, the main station system assigns corresponding weights to the charging voltage, steady-state current, transient current, and voltage-current waveforms based on the battery's capacity specifications. Based on the assigned weights, the weight parameters of the trained neural network model are adjusted accordingly. The standard charging voltage, standard steady-state current, standard transient current, and standard voltage-current waveforms of the battery under test are pre-acquired. The differences between the charging voltage, steady-state current, transient current, and voltage-current waveforms and their corresponding standard values ​​are calculated. These differences are used as input to the neural network model, and the output is the battery's lifespan data. This solution can predict the battery's lifespan based on the charging voltage, steady-state current, transient current, and voltage-current waveforms during charging. Vehicle owners can understand the vehicle's lifespan each time it is charged and take timely action.

[0121] like Figure 3 As shown, this application also provides a battery life prediction device based on a smart IoT energy meter, the device comprising:

[0122] The first acquisition module 201 is used to acquire the charging data of the battery under test during the charging period through the smart IoT energy meter. The charging data includes charging voltage, steady-state current, transient current, and voltage and current waveforms.

[0123] The local analysis module 202 is used to match charging data with the local model in the smart IoT energy meter; if the matching fails, the charging data is transmitted to the main station system.

[0124] The allocation module 203 is used to enable the master station system to configure corresponding weights for charging voltage, steady-state current, transient current, and voltage and current waveforms based on the capacity specifications of the battery under test.

[0125] The adjustment module 204 is used to adjust the weight parameters of the trained neural network model according to the assigned weights.

[0126] The second acquisition module 205 is used to acquire in advance the standard charging voltage, standard current steady-state quantity, standard current transient quantity, and standard voltage and current waveform of the battery under test.

[0127] The calculation module 206 is used to calculate the differences between the charging voltage, steady-state current, transient current, and voltage and current waveforms and the corresponding standard charging voltage, standard steady-state current, standard transient current, and standard voltage and current waveforms, respectively.

[0128] The prediction module 207 is used to take the difference as input to the neural network model and output the life data of the battery under test.

[0129] In this embodiment, the solution can be applied to a charging pile system, wherein the charging pile system includes a charging pile, a smart IoT energy meter, an energy controller, and a master station system that are connected in sequence.

[0130] Charging piles are used to connect to electric vehicles and charge them; IoT energy meters are used to record charging data during charging.

[0131] In the first acquisition module 201, charging data can be obtained through the IoT energy meter, which then transmits the charging data to the energy controller and the main station system in sequence.

[0132] In the local analysis module 202, after acquiring the charging data, local data analysis is first performed on the smart IoT energy meter. The smart IoT energy meter pre-stores some data models related to the charging data, such as pre-stored multiple sets of data. Each set of data includes charging voltage, steady-state current, transient current, voltage and current waveforms, and corresponding battery life data. If the acquired charging data matches the data model, the battery life data can be obtained directly. If they do not match, the charging data is transmitted to the main station system through the energy controller for further data analysis.

[0133] In the allocation module 203, since the charging voltage, steady-state current, transient current, and voltage-current waveforms all affect battery life, a neural network model is needed to predict battery life based on these four factors. According to their importance to the battery, the charging voltage, steady-state current, transient current, and voltage-current waveforms need to be weighted, so that factors with higher influence are assigned higher weights.

[0134] In adjustment module 204, the neural network model includes: an input layer, a convolutional layer, a fusion layer, a fully connected layer, and an output layer.

[0135] The weight parameters are located in the fully connected layer. Based on the pre-assigned weights of the charging voltage, steady-state current, transient current, and voltage and current waveforms, the weight parameters of the fully connected layer are adjusted accordingly to make the output results more accurate.

[0136] In the second acquisition module 205, the standard charging voltage, standard steady-state current, standard transient current, and standard voltage and current waveforms of the battery under test are the data when the battery is working normally or just leaving the factory. These data can be obtained by consulting relevant materials.

[0137] In calculation module 206, the current charging voltage of the battery under test is then subtracted from the standard charging voltage to obtain the voltage difference. Similarly, the steady-state current difference, transient current difference, and temperature difference can also be calculated. The purpose of this step is to enable the subsequent neural network model to integrate the current data and standard data for prediction, resulting in more accurate and universal output results.

[0138] In the prediction module 207, the difference is used as the input to the neural network model, and the output is the life data of the battery under test.

[0139] Therefore, by acquiring charging data of the battery under test during the charging period; when the charging data fails to match the local model data, the main station system assigns corresponding weights to the charging voltage, steady-state current, transient current, and voltage-current waveforms based on the capacity specifications of the battery under test; based on the assigned weights, the weight parameters of the trained neural network model are adjusted accordingly; the standard charging voltage, standard steady-state current, standard transient current, and standard voltage-current waveforms of the battery under test are pre-acquired; the differences between the charging voltage, steady-state current, transient current, and voltage-current waveforms and the corresponding standard charging voltage, standard steady-state current, standard transient current, and standard voltage-current waveforms are calculated; the differences are used as input to the neural network model, and the output yields the battery life data. This solution can predict the battery life based on the charging voltage, steady-state current, transient current, and voltage-current waveforms during charging, allowing vehicle owners to understand the vehicle's lifespan each time it is charged and take timely action.

[0140] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of those different embodiments or examples.

[0141] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0142] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for predicting battery life based on a smart IoT energy meter, characterized in that, The method includes: The charging data of the battery under test during the charging period is obtained by a smart IoT energy meter. The charging data includes charging voltage, steady-state current, transient current, and voltage and current waveforms. The smart IoT energy meter performs data matching between the charging data and the local model; if the matching fails, the charging data is transmitted to the main station system. The master station system assigns corresponding weights to the charging voltage, steady-state current, transient current, and voltage and current waveforms based on the capacity specifications of the battery under test. Based on the assigned weights, the weight parameters of the trained neural network model are adjusted accordingly. The standard charging voltage, standard steady-state current, standard transient current, and standard voltage and current waveforms of the battery under test are obtained in advance. Calculate the differences between the charging voltage, steady-state current, transient current, and voltage and current waveforms and the corresponding standard charging voltage, standard steady-state current, standard transient current, and standard voltage and current waveforms. The difference is used as the input to the neural network model, and the output is the lifespan data of the battery under test. Based on the capacity specifications of the battery under test, corresponding weights are assigned to the charging voltage, steady-state current, transient current, and voltage-current waveforms, including: If the capacity specification of the battery under test is greater than or equal to a preset threshold T, then the weight of the charging voltage is set to... The weights of the steady-state current quantities are set as follows: The weight of the transient current quantity is set as follows: The weights of the voltage and current waveforms are set as follows: Wherein, the preset threshold T is greater than 1; If the battery capacity is less than a preset threshold T, then the weight of the charging voltage is set to... The weights of the steady-state current quantities are set as follows: The weight of the transient current quantity is set as follows: The weights of the voltage and current waveforms are set as follows: ; The adjustment of the weight parameters of the trained neural network model based on the assigned weights includes: Find the smallest weight parameter among all weight parameters in the neural network model, excluding the weight parameters for charging voltage, steady-state current, transient current, and voltage / current waveforms. ; The weight parameter Q1 of the charging voltage in the neural network model ; The weight parameter Q2 of the steady-state current in the neural network model ; The weight parameter Q3 of the current transient in the neural network model ; The weighting parameter Q4 of the voltage and current waveforms in the neural network model ; Remove the weight parameters from the neural network model .

2. The method according to claim 1, characterized in that, The preset threshold T is set to 1.

5.

3. The method according to claim 1, characterized in that, After outputting the lifespan data of the battery under test, the method further includes: The neural network model in the master station system is sent to the smart IoT energy meter.

4. The method according to claim 1, characterized in that, After outputting the lifespan data of the battery under test, the method further includes: If the capacity of the battery under test is between 15 and 30 kWh, then the abnormal lifespan threshold of the battery under test is set to 50%. If the capacity of the battery under test is between 30 and 60 kWh, then the abnormal lifespan threshold of the battery under test is set to 60%. If the lifespan data is lower than the lifespan abnormality threshold, the battery under test is determined to be abnormal. If the battery under test is determined to be abnormal, the corresponding abnormality information will be notified to the vehicle owner.

5. A battery life prediction system based on a smart IoT energy meter, characterized in that, The system includes an electric vehicle charging pile, a smart IoT energy meter, an energy controller, and a master station system that are connected in sequence. During the charging process of the electric vehicle charging pile and the electric vehicle, the smart IoT energy meter acquires the charging data of the battery under test during the charging time period. The charging data includes charging voltage, steady-state current, transient current, and voltage and current waveforms. The smart IoT energy meter performs data matching between the charging data and the local model. If the matching fails, the charging data is transmitted to the main station system through the energy controller. The main station system assigns corresponding weights to the charging voltage, steady-state current, transient current, and battery temperature-voltage-current waveforms based on the capacity specifications of the battery under test. Based on the assigned weights, it adjusts the weight parameters of the trained neural network model accordingly. It pre-acquires the standard charging voltage, standard steady-state current, standard transient current, and standard battery temperature-voltage-current waveforms of the battery under test. It then calculates the differences between the charging voltage, steady-state current, transient current, and battery temperature-voltage-current waveforms and their corresponding standard values. These differences are used as input to the neural network model, and the system outputs the lifespan data of the battery under test. Based on the capacity specifications of the battery under test, corresponding weights are assigned to the charging voltage, steady-state current, transient current, and voltage-current waveforms, including: If the capacity specification of the battery under test is greater than or equal to a preset threshold T, then the weight of the charging voltage is set to... The weights of the steady-state current quantities are set as follows: The weight of the transient current quantity is set as follows: The weights of the voltage and current waveforms are set as follows: Wherein, the preset threshold T is greater than 1; If the battery capacity is less than a preset threshold T, then the weight of the charging voltage is set to... The weights of the steady-state current quantities are set as follows: The weight of the transient current quantity is set as follows: The weights of the voltage and current waveforms are set as follows: ; The adjustment of the weight parameters of the trained neural network model based on the assigned weights includes: Find the smallest weight parameter among all weight parameters in the neural network model, excluding the weight parameters for charging voltage, steady-state current, transient current, and voltage / current waveforms. ; The weight parameter Q1 of the charging voltage in the neural network model ; The weight parameter Q2 of the steady-state current in the neural network model ; The weight parameter Q3 of the current transient in the neural network model ; The weighting parameter Q4 of the voltage and current waveforms in the neural network model ; Remove the weight parameters from the neural network model .

6. The system according to claim 5, characterized in that, The master station system will also send the neural network model with adjusted weight parameters to the smart IoT energy meter through the energy controller.

7. The system according to claim 5, characterized in that, The smart IoT energy meter includes a voltage and current waveform acquisition module, a management and control unit, and a data transmission module that are connected in sequence. During the charging process of the electric vehicle charging pile and the electric vehicle, the voltage and current waveform acquisition module acquires the voltage and current waveform of the battery under test during the charging period and transmits the voltage and current waveform to the data transmission module through the management and control unit; The data transmission module transmits the voltage and current waveforms to the master station system.

8. A battery life prediction device based on a smart IoT energy meter, characterized in that, The device includes: The first acquisition module is used to acquire the charging data of the battery under test during the charging period through a smart IoT energy meter. The charging data includes charging voltage, steady-state current, transient current, and voltage and current waveforms. The local analysis module is used to perform data matching between the charging data and the local model based on the charging data in the smart IoT energy meter; if the matching fails, the charging data is transmitted to the main station system. The allocation module is used to enable the master station system to configure corresponding weights for the charging voltage, steady-state current, transient current, and voltage and current waveforms based on the capacity specifications of the battery under test. The adjustment module is used to adjust the weight parameters of the trained neural network model based on the assigned weights. Based on the capacity specifications of the battery under test, corresponding weights are assigned to the charging voltage, steady-state current, transient current, and voltage-current waveforms, including: If the capacity specification of the battery under test is greater than or equal to a preset threshold T, then the weight of the charging voltage is set to... The weights of the steady-state current quantities are set as follows: The weight of the transient current quantity is set as follows: The weights of the voltage and current waveforms are set as follows: Wherein, the preset threshold T is greater than 1; If the battery capacity is less than a preset threshold T, then the weight of the charging voltage is set to... The weights of the steady-state current quantities are set as follows: The weight of the transient current quantity is set as follows: The weights of the voltage and current waveforms are set as follows: ; The adjustment of the weight parameters of the trained neural network model based on the assigned weights includes: Find the smallest weight parameter among all weight parameters in the neural network model, excluding the weight parameters for charging voltage, steady-state current, transient current, and voltage / current waveforms. ; The weight parameter Q1 of the charging voltage in the neural network model ; The weight parameter Q2 of the steady-state current in the neural network model ; The weight parameter Q3 of the current transient in the neural network model ; The weighting parameter Q4 of the voltage and current waveforms in the neural network model ; Remove the weight parameters from the neural network model ; The second acquisition module is used to acquire in advance the standard charging voltage, standard current steady-state quantity, standard current transient quantity, and standard voltage and current waveform of the battery under test; The calculation module is used to calculate the differences between the charging voltage, steady-state current, transient current, and voltage and current waveforms and the corresponding standard charging voltage, standard steady-state current, standard transient current, and standard voltage and current waveforms, respectively. The prediction module is used to take the difference as input to the neural network model and output the life data of the battery under test.

Images