Battery pack intelligent matching method using TensorFlowAPI software and battery swap station control system

Abstract

The invention discloses a battery pack intelligent matching method using TensorFlowAPI software, and the method comprises the following steps: creating a model; acquiring parameter information of the vehicle; obtaining the bin position of the battery rack and the parameter information of the stored battery; converting the parameter information into a feature vector of a training data set; building a neural network layer; inputting the feature vector into the model, and obtaining the information of the vehicle when the vehicle is driven into the battery swap station; a matched and empty battery rack is obtained through model calculation, and the RGV robot carries the power-deficient battery to the battery rack; and the model is matched with a storage battery matched with the power-lack battery, and the RGV robot carries the storage battery on the battery rack and installs the storage battery on the vehicle. The method has the advantages that the problem that manual matching is difficult in the battery replacement process under the condition that vehicle brands and battery models are diversified is effectively solved, matching of parameters such as the size, the weight, the voltage and the current of the battery is achieved, and the working efficiency is greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of battery swapping, and in particular, to a method for intelligent matching of battery packs using TensorFlow API software and a control system for a battery swapping station. Background Art

[0002] With the rapid development of new energy vehicles, battery swapping stations have become more and more popular. A battery swapping station is a place that provides battery swapping services for new energy vehicles, which can quickly replace the battery of new energy vehicles and save charging time. The operating efficiency of a battery swapping station depends on the matching degree of factors such as vehicles, batteries, and battery rack positions for storing batteries. At present, the matching method of battery swapping stations mainly relies on manual or database program matching: manual matching is carried out according to factors such as vehicle models and battery capacities. Due to the large number of matching factors and the rather complex scenarios, the efficiency of manual or database matching is low, prone to errors, and increases potential safety hazards. Summary of the Invention

[0003] The purpose of the present invention is to overcome the deficiencies of the prior art and provide a method for intelligent matching of battery packs using TensorFlow API software and a control system for a battery swapping station.

[0004] The purpose of the present invention is achieved through the following technical solutions:

[0005] A method for intelligent matching of battery packs using TensorFlow API software includes the following steps:

[0006] S1. Collect the corresponding data of vehicle batteries supported by the battery swapping station, the battery racks, and the data parameters of the batteries stored on the battery racks as the training data set, and create a model using TensorFlow software;

[0007] S2. Obtain the parameters of the vehicle and the parameter information of the discharged battery in the vehicle;

[0008] S3. Obtain the positions of the battery racks and the parameter information of the stored batteries;

[0009] S4. Convert the above parameter information into the feature vectors of the training data set through the concatenate method of NumPy;

[0010] S5. Obtain the features of the data set to build a neural network layer;

[0011] S6. Input the feature vectors into the model, and start the training of the matching model using the model.fit method in TensorFlow API until the performance of the model meets the requirements;

[0012] S7. When the vehicle drives into the battery swapping station, obtain the information of the vehicle;

[0013] S8. The model calculates a suitable and empty battery rack, and the RGV robot transports the discharged battery to the battery rack.

[0014] S9. The model matches a stored battery suitable for the discharged battery, and the RGV robot transports and installs the stored battery on the battery rack to the vehicle.

[0015] Preferably, in step S2, "obtaining the parameters of the vehicle and the parameter information of the discharged battery in the vehicle" specifically means: obtaining the parameters of the vehicle's brand, model, battery model, capacity, material, size, charging voltage, and charging current, and storing them in the vehicle_info.csv file. Each row of the csv file represents a vehicle, and each column represents a parameter; loading through np.loadtxt called by NumPy; preprocessing the data through tf.data of the TensorFlow API.

[0016] Preferably, in step S3, "obtaining the positions of the battery rack and the parameter information of the stored battery" specifically means: obtaining the model of the battery rack, the battery rack number, the battery model, capacity, charging voltage, the length, width, and height of the battery rack position, and the parameter of whether it supports charging; storing the corresponding battery rack information data in the warehouse_info.csv file, loading through np.loadtxt in NumPy; preprocessing the data through tf.data of the TensorFlow API.

[0017] Preferably, the "feature vector in step S4" specifically includes vehicle information, the position information of the battery rack, the information of the discharged battery, and the parameter information of the stored battery.

[0018] Preferably, "S5. Obtaining the features of the dataset to build a neural network layer" specifically means: the features of the dataset include the model of the vehicle, the model of the battery, the model of the battery rack, the capacity, voltage, and current information of the battery, and building the following neural network layer according to the above information:

[0019] model = tf.keras.Sequential(

[0020] tf.keras.layers.Dense(128, activation='relu'),

[0021] tf.keras.layers.Dense(64, activation='relu'),

[0022] tf.keras.layers.Dense(4, activation='softmax')])。

[0023] Preferably, "Step S7: When the vehicle enters the battery swapping station, obtain the information of the vehicle;" specifically: The T-BOX on the vehicle communicates with the vehicle's battery system through the CAN bus to obtain vehicle battery information, including information parameters such as battery specifications, battery power, battery temperature, and battery status.

[0024] Preferably, "the model calculates the adapted and empty battery rack" specifically: Load the model through tf.keras.models.load_model, input data through the method output_data = model, and obtain the output result of the model, that is, the position of the target battery rack where the undercharged battery needs to be placed, through the method prediction = output_data.numpy.

[0025] Preferably, the stored battery is a battery pack with a power of 50% - 100%.

[0026] The battery swapping station control system includes a memory, a processor, and a battery swapping instruction control system stored on the memory and operable on the processor. When the battery swapping instruction control system performs battery swapping, it implements the battery pack intelligent matching method using the TensorFlow API software as described in any one of the above.

[0027] The beneficial effects of the present invention are mainly reflected in:

[0028] 1. Exquisite design, effectively solving the problem of difficult manual matching during the battery swapping process in the case of diverse vehicle brands and battery models, realizing the matching of parameters such as the size, weight, voltage, and current of the battery, facilitating automatic battery swapping for different brands of vehicles at the battery swapping station, with a high degree of automation and greatly improving work efficiency;

[0029] 2. The model created by TensorFlow software will continuously learn to extract features from the feature vectors and classify or predict the features, thereby realizing the full-automatic matching of the battery pack, avoiding the occurrence of manual misoperations, ensuring the accuracy rate, and reducing potential safety hazards;

[0030] 3. The battery pack intelligent matching method using the TensorFlow API software adopts a deep learning algorithm, which can learn the relationship between the parameters of the battery, vehicle, and battery rack, thereby more accurately evaluating the matching degree of the battery pack;

[0031] 4. Using this method can utilize the GPU for parallel computing, thereby improving the matching efficiency. In addition, it can also be flexibly adjusted according to the parameters of the battery pack to adapt to different types of battery packs.

[0032] 5. This method can learn the relationships between the parameters of the battery pack, battery rack, and vehicle, so as to more accurately evaluate the matching degree of the battery pack. In addition, the TensorFlow API software can ensure the safety of the battery pack, give an alarm for the matching degree data, and maximize the improvement of safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] The technical solution of the present invention will be further described below in conjunction with the drawings:

[0034] Figure 1 : Schematic diagram of the preferred embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0035] The present invention will be described in detail below in conjunction with the specific embodiments shown in the drawings. However, these embodiments are not limited to the present invention, and any structural, method, or functional transformation made by those of ordinary skill in the art based on these embodiments is included in the protection scope of the present invention.

[0036] The present invention will be described in detail below with reference to the drawings and in conjunction with embodiments.

[0037] As Figure 1 shown, the invention discloses an intelligent matching method for battery packs using TensorFlow API software, including the following steps.

[0038] S1. Collect the data parameters corresponding to the vehicle batteries supported by the battery swapping station, the battery racks, and the storage batteries stored on the battery racks as the training data set, and use the TensorFlow software to create a model. Specifically, the following information needs to be collected: the model of the vehicle, the model of the battery, the model of the battery rack, the capacity, voltage, and current of the battery. The size, weight, etc. of the battery rack.

[0039] The content of the vehicle training data set in tabular format is as follows:

[0040]

[0041] The content of the battery rack training data set in tabular format is as follows:

[0042]

[0043]

[0044] S2. Obtain the parameters of the vehicle and the parameter information of the discharged battery in the vehicle. Specifically, obtain the brand and model of the vehicle, and the model, capacity, material, size, charging voltage, charging current, etc. of the battery, and store them in the vehicle_info.csv file. Each row of the csv file represents a vehicle, and each column represents a parameter. Load it by calling np.loadtxt("vehicle_info.csv", delimiter = "," ) of NumPy. Preprocess the data through tf.data of the TensorFlow API.

[0045] S3. Obtain the bin positions of the battery rack and the parameter information of the stored batteries. Specifically, obtain the model of the battery rack, the battery rack number, the model, capacity, charging voltage, the length, width and height of the battery rack bin, whether it supports charging and other parameters. Store the corresponding battery rack information data in the warehouse_info.csv file and load it by calling np.loadtxt("warehouse_info.csv", delimiter = "," ) in NumPy. Preprocess the data through tf.data of the TensorFlow API.

[0046] S4. According to the matching of the parameters of the discharged battery on the vehicle and the parameters of the batteries in the battery rack, convert these parameters into the feature vector features of the training dataset of the TensorFlow artificial intelligence framework through the concatenate method of NumPy (NumPy is the basic software package for scientific computing in Python), that is, features = np.concatenate((vehicle_info, warehouse_info), axis = 1). The feature vector is a vector that contains vehicle information, battery rack bin information and discharged battery information parameters. The dimension of the feature vector is determined by the number of discharged battery information, vehicle information, battery rack bin information and battery information parameters.

[0047] S5. Obtain the features of the dataset to build the neural network layer. Specifically, the construction of the neural network layer needs to be determined according to the features of the dataset. For the matching problem of vehicle batteries and battery rack bins in the battery swapping station, the features of the dataset include the model of the vehicle, the model of the battery, the model of the battery rack, the capacity, voltage, current, etc. of the battery. Therefore, the neural network layer can be built as follows:

[0048] model = tf.keras.Sequential(

[0049] tf.keras.layers.Dense(128, activation='relu'),

[0050] tf.keras.layers.Dense(64, activation='relu'),

[0051] tf.keras.layers.Dense(4, activation='softmax')])。

[0052] S6. Input the feature vector into the model, and use the model.fit(features, labels, epochs = 10) method in the TensorFlow API to start the training of the matching model. Through the backpropagation algorithm, the TensorFlow model will continuously learn to extract features from the feature vector and classify or predict the features. When the performance of the model meets the requirements, the training process ends. In the above, when the model meets the requirements, it means that a storage battery suitable for the discharged battery can be matched. Of course, other conditions can also be added according to needs, which all belong to the protection scope of the present invention and will not be elaborated here.

[0053] S7. When the vehicle enters the battery swapping station, obtain the information of the vehicle. Specifically, the vehicle T-BOX communicates with the battery system of the vehicle through the CAN bus to obtain the vehicle battery information, including battery power, battery temperature, battery status, etc. At the same time, the in-vehicle T-BOX establishes wireless communication with the battery swapping station to transmit the battery information of the vehicle. In addition, the model, battery material (ternary lithium battery, lithium iron phosphate, etc.), manufacturer, length, width, and height dimensions, allowable charging voltage, charging current magnitude, and whether there is a battery on the battery rack of the existing battery pack in the battery rack warehouse can be obtained.

[0054] S8. The model calculates the suitable and empty battery rack, and the RGV robot transports the discharged battery to the battery rack. Specifically, when the vehicle arrives at the battery swapping station, the battery parameter information of the current vehicle is obtained through the station control system of the battery swapping station. The vehicle battery parameter information is read through the system and input into the trained and evaluated model to calculate a suitable empty battery rack, and the discharged battery on the vehicle is transferred to the battery rack. The specific method is through

[0055] tf.keras.models.load_model("model.h5") to load the model, and through the method output_data =

[0056] The model(input_data) takes the input data, and the output result of the model, which is the position of the target battery rack where the depleted battery needs to be placed, is obtained through the method prediction = output_data.numpy(). The control system of the battery swapping station matches the vehicle with the available battery racks in the station according to the battery information transmitted by the vehicle's T-BOX. After successful matching, the control system of the battery swapping station gives a battery swapping instruction. The depleted battery is taken out of the vehicle and will automatically fall onto the matching empty battery rack in the station.

[0057] S9. The model matches the storage battery suitable for the depleted battery, and the RGV robot transports and installs the storage battery on the battery rack onto the vehicle. In the above, a trained and evaluated model is used to obtain an output vehicle model information, and it is compared whether the vehicle model output by the system model is the same as the current vehicle. If it is a match, after obtaining the battery data, the battery rack position where the battery information is located is found. At this time, there may be multiple battery positions that meet the battery swapping requirements, and we select the battery with the highest power among them to complete the matching. If the battery is charging, it is matched after real-time reading by the system. If no battery that meets the battery swapping requirements is found, an error message indicating that none has been found is returned to the station control system. The control system of the battery swapping station will command the RGV robot to take out the fully charged battery on the battery rack. The fully charged battery will automatically move under the vehicle waiting. Additionally, in the above, the storage battery is a battery pack with a power of 50% - 100%.

[0058] In the present invention, a comparison is made with the prior art in terms of data source and data processing method:

[0059] The data sources of the traditional database matching method and the manual matching method are static index data such as the specification parameters of the battery pack, while the data source of the intelligent battery pack matching method using the TensorFlow API software can be the real-time operation data of the battery pack. The operation data includes information such as the temperature, voltage, current, and SOC of the battery pack, and these information can more comprehensively reflect the performance of the battery pack, such as the remaining power information.

[0060] Both the traditional database matching method and the manual matching method are based on the specification parameters of the battery pack for matching, while the intelligent battery pack matching method using the TensorFlow API software adopts a deep learning algorithm, which can learn the relationships between the various parameters of the battery, vehicle, and battery rack, so as to more accurately evaluate the matching degree of the battery pack.

[0061] A comparison is made with the prior art from the perspective of the matching algorithm:

[0062] Traditional database matching methods can only match based on the specification parameters of the battery pack, while manual matching methods require manual experience judgment, both of which have certain errors. The intelligent battery pack matching method using TensorFlow API software adopts deep learning algorithms and can learn the matching rules of the battery pack, thereby improving the matching accuracy. Traditional database matching methods use simple matching algorithms such as hash algorithms and Boolean algorithms. These algorithms can only match based on the specification parameters of the battery pack and cannot consider the operating status of the battery pack. However, the intelligent battery pack matching method using TensorFlow API software adopts deep learning algorithms and can learn dynamic data such as the operating data of the battery pack, the specifications of the battery rack, and whether there are batteries, so as to more accurately evaluate the matching degree of the battery pack.

[0063] Compare with the prior art from the perspective of matching efficiency:

[0064] Traditional database matching methods need to query database records one by one, with low efficiency. Manual matching methods require manual judgment one by one, with even lower efficiency. The intelligent battery pack matching method using TensorFlow API software can utilize GPU for parallel computing, thereby improving the matching efficiency.

[0065] Compare with the prior art from the perspective of automation:

[0066] Both traditional database matching methods and manual matching methods require manual operations, while the intelligent battery pack matching method using TensorFlow API software can automate the battery pack matching, thereby reducing the burden of manual operations.

[0067] Compare with the prior art from the perspective of scalability:

[0068] Traditional database matching methods need to modify and adjust the matching rules in the database, while manual matching methods require manual calculation of the rules, both of which have certain limitations. The intelligent battery pack matching method using TensorFlow API software can be flexibly adjusted according to the parameters of the battery pack, so as to adapt to different types of battery packs.

[0069] Compare with the prior art from the perspective of cost and accuracy:

[0070] The usage cost of the TensorFlow API software is relatively low, thus reducing the cost of battery pack matching. Based on the above technical features, the battery pack intelligent matching method using the TensorFlow API software has the following specific technical effects: The deep learning algorithm can learn the relationships between the parameters of the battery pack, the battery rack, and the vehicle, so as to more accurately evaluate the matching degree of the battery pack. In addition, the TensorFlow API software can achieve the safety guarantee of the battery pack and give an alarm for the matching degree data. For example, when the matching degree is lower than 99.9%, it gives information for manual judgment to prompt manual intervention, thus avoiding safety accidents of the battery pack. The traditional database matching method and the manual matching method cannot achieve safety guarantee.

[0071] The present invention also discloses a swapping station control system, including a memory, a processor, and a swapping instruction control system stored on the memory and operable on the processor. When the swapping instruction control system executes swapping, it implements the battery pack intelligent matching method using the TensorFlow API software as described in any one of the above.

[0072] It should be understood that although this specification is described according to the embodiments, not every embodiment only contains an independent technical solution. This narrative way of the specification is only for clarity. Those skilled in the art should regard the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

[0073] The series of detailed descriptions listed above are only specific descriptions of the feasible embodiments of the present invention, and they are not used to limit the protection scope of the present invention. Any equivalent embodiments or changes made without departing from the technical spirit of the present invention should be included in the protection scope of the present invention.

Claims

1. A battery pack intelligent matching method using the TensorFlow API software, comprising the following steps: Characterized in that: S1. Collect the corresponding data of vehicle batteries supported by the battery swapping station, the battery racks, and the data parameters of the stored batteries placed on the battery racks as the training dataset, and create a model using TensorFlow software; S2. Obtain the parameters of the vehicle and the parameter information of the discharged battery in the vehicle; S3. Obtain the bin positions of the battery racks and the parameter information of the stored batteries; S4. Convert the above parameter information into the feature vectors of the training dataset through the concatenate method of NumPy; S5. Obtain the features of the dataset to build the neural network layer; S6. Input the feature vectors into the model, and start the training of the matching model using the model.fit method in the TensorFlow API until the performance of the model meets the requirements; S7. When the vehicle enters the battery swapping station, obtain the information of the vehicle; S8. The model calculates the suitable and empty battery racks, and the RGV robot transports the discharged battery to the battery racks; S9. The model matches the stored battery suitable for the discharged battery, and the RGV robot transports and installs the stored battery on the battery racks to the vehicle.

2. The battery pack intelligent matching method using the TensorFlow API software according to claim 1, Characterized in that: In step S2, "obtain the parameters of the vehicle and the parameter information of the discharged battery in the vehicle" is specifically: obtain the parameters of the vehicle brand, model, battery model, capacity, material, size, charging voltage, and charging current, and store them in the vehicle_info.csv file. Each row of the csv file represents a vehicle, and each column represents a parameter; load through the np.loadtxt of NumPy; preprocess the data through tf.data of the TensorFlow API.

3. The battery pack intelligent matching method using the TensorFlow API software according to claim 1, Characterized in that: In step S3, "obtain the bin positions of the battery racks and the parameter information of the stored batteries" is specifically: obtain the parameters of the battery rack model, battery rack number, battery model, capacity, charging voltage, length, width, height of the battery rack bin, and whether charging is supported; store the corresponding battery rack information data in the warehouse_info.csv file, and load through the np.loadtxt in NumPy; Preprocess the data through tf.data of the TensorFlow API.

4. The battery pack intelligent matching method using the TensorFlow API software according to claim 1, Characterized in that: The "feature vectors in step S4" specifically include vehicle information, bin position information of the battery racks, discharged battery information, and stored battery information parameters.

5. The battery pack intelligent matching method using the TensorFlow API software according to claim 1, Characterized in that: "S5. Obtain the features of the dataset to build the neural network layer" specifically means: The features of the dataset include the vehicle model, battery model, battery rack model, battery capacity, voltage, and current information, and the following neural network layer is built based on the above information: model = tf.keras.Sequential( tf.keras.layers.Dense(128, activation='relu'), tf.keras.layers.Dense(64, activation='relu'), tf.keras.layers.Dense(4, activation='softmax')])。 6. The battery pack intelligent matching method using the TensorFlow API software according to claim 1, characterized in that: "Step S7. When the vehicle enters the battery swapping station, obtain the vehicle information;" specifically means: The T-BOX on the vehicle communicates with the vehicle's battery system through the CAN bus to obtain vehicle battery information, including information parameters such as battery specifications, battery power, battery temperature, and battery status.

7. The battery pack intelligent matching method using the TensorFlow API software according to claim 1, characterized in that: "The model calculates the suitable and empty battery rack" specifically means: Load the model through tf.keras.models.load_model, input data through the method output_data = model, and obtain the output result of the model, that is, the position of the target battery rack where the discharged battery needs to be placed, through the method prediction = output_data.numpy.

8. The battery pack intelligent matching method using the TensorFlow API software according to claim 1, characterized in that: The stored battery is a battery pack with a power of 50% to 100%.

9. The battery swapping station control system, characterized in that: It includes a memory, a processor, and a battery swapping instruction control system stored on the memory and operable on the processor. When the battery swapping instruction control system executes battery swapping, it implements the battery pack intelligent matching method using the TensorFlow API software as described in any one of claims 1-8.

Images