Battery surface temperature calculation method and system and battery storage device
Through the neural network model combining environmental and battery characteristic data, the reading of the non-contact temperature sensor is corrected, which solves the problem of inaccurate battery surface temperature measurement in three-dimensional storage and achieves more accurate battery status monitoring.
Patent Information
- Application Number
- CN202411296002.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-14
- Publication Date
- 2024-12-27
AI Technical Summary
In a three-dimensional storage environment, existing non-contact temperature sensors are difficult to accurately measure the surface temperature of the battery, making it difficult to detect abnormal battery states in a timely manner.
By obtaining multiple sets of training temperature and historical feature data on the battery surface, a temperature calculation model is obtained based on neural network training, and the real-time environment and battery feature data are calculated. The readings of the non-contact temperature sensor are corrected to improve measurement accuracy.
Accurate measurement of battery surface temperature is achieved, false alarms and errors are reduced, and the accuracy of battery status monitoring is improved.
Smart Images
Figure CN119203755A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of battery storage, and specifically provides a battery surface temperature calculation method and system and a battery storage device. Background Art
[0002] Vertical warehousing is a modern warehousing method that maximizes the use of vertical space and uses high-bay shelves to effectively increase the storage capacity of warehouse goods, reduce floor space, and reduce operating costs. At the same time, it uses automated equipment and systems to improve the speed and accuracy of cargo handling and work efficiency. With the increasing number of electric vehicle users, the scale of existing batteries is getting larger and larger, and the layout of battery replacement stations is becoming wider and wider. Vertical warehousing of electric vehicle batteries is in line with the current development trend.
[0003] Modern warehouse safety systems generally use gas, humidity and temperature sensors to collect relevant information. These information dimensions have their own advantages and disadvantages. Gas sensors are used to detect the concentration of flammable and toxic gases in the area. When the concentration exceeds the set threshold, an alarm will be triggered. However, gas has a diffuse characteristic, and it is difficult to accurately locate the source of the accident, and there may be false alarms when the air circulation is not smooth. Humidity sensors are used to detect changes in humidity in the area. Once the humidity exceeds the set threshold, an alarm will be triggered. However, in the electric vehicle battery storage scenario, although humidity is generally highly coupled with temperature, it is not directly related to abnormal battery conditions. Temperature sensors include contact temperature sensors and non-contact temperature sensors. Contact sensors directly contact the battery for temperature measurement, such as thermocouple sensors and thermistor sensors. They are usually low-cost, fast response and high-precision. Non-contact temperature sensors, such as infrared temperature sensors, measure temperature by measuring the infrared radiation emitted by objects. They are suitable for long-distance measurement and high-temperature measurement. For highly structured warehousing scenarios, non-contact temperature sensors are obviously more suitable for the needs of security systems. However, non-contact temperature sensors are often greatly affected by environmental data and surface characteristics of objects. In other words, the surface temperature of the battery obtained in a storage environment solely by relying on non-contact temperature sensors is inaccurate, which makes it difficult to detect abnormal battery conditions in a timely manner.
[0004] Accordingly, the art needs a new technical solution to solve the above problems. Summary of the invention
[0005] The present invention aims to solve the above technical problem, that is, to solve the problem of inaccurate battery temperature measurement in existing three-dimensional warehouses.
[0006] In a first aspect, the present invention provides a method for calculating the surface temperature of a battery, the method comprising: obtaining multiple sets of training temperatures and multiple sets of historical characteristic data of the battery surface, wherein the training temperatures and the historical characteristic data correspond one to one, and each set of historical characteristic data includes environmental data and battery data; performing neural network training based on the training temperatures and the historical characteristic data corresponding to the training temperatures to obtain a temperature calculation model; obtaining real-time environmental characteristic data of the space where the battery is located and real-time characteristic data of the battery; and inputting the real-time characteristic data into the temperature calculation model to calculate and obtain the real-time battery surface temperature.
[0007] In a preferred technical solution of the above calculation method, the environmental data includes ambient temperature and / or ambient humidity and / or light intensity and / or wind speed.
[0008] In the preferred technical solution of the above calculation method, the battery data includes battery material and / or battery color and / or battery storage location and / or battery cleanliness and / or measured battery surface temperature.
[0009] In the preferred technical solution of the above calculation method, the temperature calculation model includes multiple hidden layers, each hidden layer includes multiple neurons, and the input of each neuron is Output is H k,i =φ(G k,i ), After K hidden layers, the output result is y; where G k,i represents the input of the i-th neuron in the K-th hidden layer, is the weight of the jth input data in the ith neuron in the kth hidden layer, x j is the jth input data, represents the input bias term of the i-th neuron in the k-th layer; n k-1 Indicates the number of input data in the kth layer.
[0010] In a second aspect, the present invention also provides a battery surface temperature calculation system, which includes: an environmental data acquisition module: capable of acquiring environmental data; a battery data acquisition module: capable of acquiring battery data; a processor module: based on the acquired environmental data and battery data, a temperature calculation model is used to calculate the real-time battery surface temperature.
[0011] In the preferred technical solution of the above-mentioned computing system, the battery data acquisition module includes a non-contact temperature sensor, which can directly measure the battery surface temperature before correction; the processor module calculates the corrected real-time battery surface temperature based on the battery surface temperature and environmental data before correction using a temperature calculation model.
[0012] In a preferred technical solution of the above-mentioned computing system, the battery data acquisition module includes a scanning device, an identification image is provided on the battery, and the scanning device can obtain the battery material and / or battery color after scanning the identification image.
[0013] In the preferred technical solution of the above-mentioned computing system, the computing system also includes a cloud module; the cloud module is capable of receiving information acquired by the environmental data acquisition module and the battery data acquisition module, and transmitting the received information to the processor module.
[0014] In the preferred technical solution of the above computing system, the environmental data acquisition module includes: an environmental temperature sensor and / or an environmental humidity sensor and / or a light intensity sensor and / or a wind speed sensor.
[0015] In a third aspect, the present invention further provides a battery storage device, which includes a battery surface temperature calculation system as described in any one of the above-mentioned preferred technical solutions, and is capable of executing a battery surface temperature calculation method as described in any one of the above-mentioned preferred technical solutions.
[0016] In the case of adopting the above technical solution, the battery surface temperature calculation method of the present invention includes: obtaining multiple sets of training temperatures and multiple sets of historical feature data of the battery surface, wherein the training temperatures and the historical feature data correspond one to one, and each set of historical feature data includes environmental data and battery data; performing neural network training based on the training temperatures and the historical feature data corresponding to the training temperatures to obtain a temperature calculation model; obtaining real-time environmental feature data of the space where the battery is located and real-time feature data of the battery; inputting the real-time feature data into the temperature calculation model to calculate and obtain the real-time battery surface temperature. The present invention calculates the battery surface temperature through battery data and environmental data, and can accurately measure the temperature of the battery surface. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings, in which:
[0018] Figure 1 It is a flow chart of the main steps of the battery surface temperature calculation method of the present invention;
[0019] Figure 2 It is a comparison chart of normal and distorted imaging;
[0020] Figure 3 is the relationship between the sensitivity of the infrared temperature sensor and the angle of incidence;
[0021] Figure 4 It is a schematic diagram of the deep neural network structure of multivariate functions. DETAILED DESCRIPTION
[0022] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the scope of protection of the present invention. Those skilled in the art can make adjustments to them as needed to adapt to specific application scenarios.
[0023] It should be noted that in the description of the present invention, "module" may include hardware, software or a combination of both. A module may include hardware circuits, various suitable sensors, communication ports, and memories, and may also include software parts, such as program codes, or a combination of software and hardware. The term "A and / or B" means all possible combinations of A and B, such as only A, only B, or A and B.
[0024] When using non-contact temperature sensors to measure the surface temperature of batteries, such as infrared temperature sensors, they usually have the disadvantages of optical path distortion and attenuation as the incident angle increases. Optical path distortion refers to the imaging distortion caused by different lens diopter and aperture position in different fields of view in an optical system. This distortion will affect the accuracy of imaging and bring errors to analysis and measurement. Common distortions are divided into pincushion distortion and barrel distortion. Their imaging characteristics can be seen in Figure 2 . Figure 2 In the figure, a represents normal imaging, b represents pincushion distortion, and c represents barrel distortion. The image formed by pincushion distortion b will appear to be concave in the center and convex at the edges, which is due to the different refractive indices between the central and peripheral areas of the lens. When light enters from the center of the lens, the refractive index in the center of the lens is higher, the light is bent to a greater extent, and the image formed is smaller than the original object; while when light enters from the edge of the lens, the refractive index at the edge of the lens is lower, the light is bent to a lesser extent, and the image formed is magnified than the original object. Barrel distortion c is the opposite of pincushion distortion b. The image formed will appear to be convex in the center and concave at the edges. When light enters from the center of the lens, the refractive index in the center of the lens is lower, the light is bent to a lesser extent, and the image formed is magnified than the original object; while when light enters from the edge of the lens, the refractive index at the edge of the lens is higher, the light is bent to a greater extent, and the image formed is smaller than the original object.
[0025] The intensity of infrared light received by the infrared temperature sensor is related to the incident angle. Its characteristics are as follows: Figure 3As shown. When the incident angle of infrared light increases, the incident light will be closer to parallel with the normal direction perpendicular to the sensor surface, which will cause the distance the light travels in the medium to increase. Since infrared light will be scattered and absorbed during its propagation in the medium, the intensity of the light will gradually decay as the propagation distance increases. Therefore, when the incident angle increases, the intensity of infrared light received by the sensor will decrease, causing the sensor reading to deviate from the actual situation. Factors that affect the sensor reading also include environmental factors and battery factors. Therefore, it is not accurate enough to use a non-contact temperature sensor to measure the battery surface temperature.
[0026] In order to solve the above problems, the present invention provides a method for calculating the surface temperature of a battery. Figure 1 , the calculation method includes:
[0027] Step S1: Acquire multiple sets of training temperatures and multiple sets of historical characteristic data of the battery surface;
[0028] Step S2: performing neural network training based on the training temperature and the historical characteristic data corresponding to the training temperature to obtain a temperature calculation model;
[0029] Step S3: Acquire real-time environmental characteristic data of the space where the battery is located and real-time characteristic data of the battery;
[0030] Step S4: input the real-time characteristic data into the temperature calculation model to calculate and obtain the real-time battery surface temperature.
[0031] In step S1, the training temperature is the measured temperature of the battery surface obtained by a contact temperature sensor, that is, the temperature sensor directly contacts the battery surface, and the measured temperature, multiple groups of historical feature data include battery data and environmental data, the battery data includes battery material and / or battery color and / or battery storage location and / or battery cleanliness and / or battery surface temperature measured by a non-contact temperature sensor, and the environmental data includes ambient temperature and / or ambient humidity and / or light intensity and / or wind speed.
[0032] In step S2, the battery data and environmental data are preprocessed and standardized according to specific rules, and then all the data are used as input information for neural network training. The temperature measured by the contact temperature sensor is used as the final output information of the neural network training. The battery data and environmental data correspond to the training temperature one by one. After training, a temperature calculation model is obtained. The temperature calculation model is a neural network model. After inputting the battery data and environmental data, an output value corresponding to the current battery data and environmental data can be accurately obtained.
[0033] In step S3, the real-time environmental characteristic data of the space where the battery is located and the real-time characteristic data of the battery are obtained. It should be noted that the obtained real-time environmental characteristic data and the real-time characteristic data of the battery are preprocessed data, which are the same as the preprocessing rules used when training the model in step S2. Then step S4 is executed. In step S4, the real-time environmental characteristic data and the real-time characteristic data of the battery standardized according to the preprocessing rules of step S3 are input, so that the real-time battery surface temperature can be accurately calculated.
[0034] In this preferred embodiment, the battery data includes battery material, battery color, battery storage location, battery cleanliness and battery surface temperature measured by a non-contact temperature sensor. Of course, this is not restrictive. In other embodiments, the battery data includes any one or any combination of battery material, battery color, battery storage location, battery cleanliness and battery surface temperature measured by a non-contact temperature sensor.
[0035] The battery surface temperature measured by the non-contact temperature sensor may be affected by environmental factors or battery surface factors, resulting in inaccurate readings, and thus inaccurate battery surface temperature measurements. Therefore, when training the temperature calculation model, environmental factors and battery surface factors are taken into consideration, and the current non-corrected readings and environmental factors or other battery factors are input into the temperature calculation model to correct the readings of the non-contact temperature sensor to make them more accurate.
[0036] In this preferred example, the environmental data includes ambient temperature, ambient humidity, light intensity and wind speed. Of course, this is not restrictive. In other embodiments, the environmental data includes any one or any combination of ambient temperature, ambient humidity, light intensity and wind speed.
[0037] In this preferred embodiment, when training the neural network temperature calculation model, the output data is the battery surface temperature directly measured by the contact temperature sensor at the current moment, and the input data is the battery material, battery color, battery storage location, battery cleanliness, battery surface temperature measured by a non-contact temperature sensor, ambient temperature, ambient humidity, light intensity and wind speed corresponding to the directly measured battery surface temperature at the current moment. The above data is a group of data. By measuring the data at different times, multiple groups of input and output data are obtained, and a temperature calculation model can be trained based on a neural network.
[0038] The temperature calculation model includes multiple hidden layers, each hidden layer includes multiple neurons, and the input of each neuron is Output is H k,i =φ(G k,i ), After K hidden layers, the output result is y; where Gk,i represents the input of the i-th neuron in the K-th hidden layer, is the weight of the jth input data in the ith neuron in the kth hidden layer, x j is the jth input data, represents the input bias term of the i-th neuron in the k-th layer; n k-1 Indicates the number of input data in the kth layer.
[0039] Specifically, refer to 4. Assume that the input layer has n0 input data, that is, the number of input data of hidden layer 1 is n0, which are The input from the input layer to the i-th neuron in hidden layer 1 is: The output of the i-th neuron is H 1,i =φ(G 1,i ).
[0040] The number of input data of hidden layer 2 depends on the number of neurons in hidden layer 1. Assuming that the number of neurons in hidden layer 1 is n1, the number of input data of hidden layer 2 is n1, which are x1=H 11 , x2=H 12 , Then the input of the i-th neuron from hidden layer 1 to hidden layer 2 is: The output of the i-th neuron is H 2,i =φ(G 2,i ).
[0041] The number of input data of hidden layer 3 depends on the number of neurons in hidden layer 2. Assuming that the number of neurons in hidden layer 2 is n2, the number of input data of hidden layer 3 is n2, which are x1=H 21 , x2=H 22 , Then the input of the i-th neuron from hidden layer 1 to hidden layer 2 is: The output of the i-th neuron is H 3,i =φ(G 3,i ).
[0042] By analogy, the number of input data of hidden layer K depends on the number of neurons in hidden layer K-1. Assume that the number of neurons in hidden layer K-1 is n. k-1 , respectively The input to the i-th neuron is The output of the i-th neuron is H k,i =φ(G k,i ).
[0043] The number of input data of the output layer Y depends on the number of neurons in the hidden layer K. Assume that the number of neurons in the hidden layer K is n k , respectively x1=H K1 , x2=H K2 ,
[0044]
[0045] Furthermore, H k,i =φ(G k,i ), That is, the input G of a neuron k,i If it is greater than 0, then the output data H k,i =G k,i By itself, if a neuron input G k,i If it is less than or equal to 0, then the output data H k,i = 0. Of course, this is not restrictive. In other embodiments, the output data of a neuron is the input function or The calculated data, G k-1,i Represents the data input from multiple neurons in the previous hidden layer to this hidden layer.
[0046] It should be noted that there are i neurons in each hidden layer, and the number of neurons in each hidden layer may be different or the same. Although i is used to represent the i-th neuron in the above text, this does not mean that the number of neurons in each hidden layer is the same. In other words, the maximum value of i in no two hidden layers may be different, and technical personnel in this field can set it according to actual conditions.
[0047] In addition, it should be noted that the present invention does not impose any limitation on the number of hidden layers. The number of hidden layers can be 1, 2, 3, or any other number of layers, as long as the temperature corresponding to the input data can be accurately calculated in the end. Those skilled in the art can set it according to actual conditions.
[0048] In a second aspect, the present invention also claims protection for a battery surface temperature calculation system, the calculation system comprising an environmental data acquisition module, a battery data acquisition module and a processor module, the environmental data acquisition module being capable of acquiring environmental data, the acquired environmental data including ambient temperature and / or ambient humidity and / or light intensity and / or wind speed; and also comprising a battery data acquisition module, the battery data acquisition module being capable of acquiring battery data; the acquired battery data including battery material and / or battery color and / or battery storage location and / or battery cleanliness and / or battery surface temperature measured using a non-contact temperature sensor; the processor module being capable of calculating the real-time battery surface temperature using a temperature calculation model based on the acquired environmental data and battery data.
[0049] Furthermore, the battery data acquisition module includes a non-contact temperature sensor, which can directly measure the battery surface temperature before correction; the processor module calculates the corrected real-time battery surface temperature based on the battery surface temperature and environmental data before correction using a temperature calculation model.
[0050] In this preferred embodiment, the battery data acquisition module also includes a scanning device. An identification image is set on the battery. The scanning device can obtain the battery material and / or battery color after scanning the identification image. The identification image can be a QR code, a barcode or a specific pattern. These images are designed to contain specific information about the battery. Once the identification image is scanned, the scanning device needs to decode the data in the image. The image is converted into digital information, and then the information is parsed to obtain the properties of the battery. The parsed data contains information such as the material and color of the battery.
[0051] After the material and color are identified, different materials are recorded as different values, and different colors are recorded as different values.
[0052] The computing system also includes a camera that can capture images of the battery surface, and then perform image processing and analysis on the captured images to evaluate the neatness of the battery surface, that is, the cleanliness. The cleanliness can be divided into levels, and each level corresponds to a different value.
[0053] The battery storage position is the coordinate of the non-contact temperature sensor relative to the center of the battery. A global coordinate system can be defined. For example, in a battery storage room, the battery storage room includes multiple battery storage spaces (which can be storage racks or storage warehouses). With a corner of the battery storage room as the world coordinate origin, the coordinates of the center of each battery storage space can be measured, and the coordinates of the installed non-contact temperature sensor can also be measured. The coordinates of the non-contact temperature sensor relative to the center of the battery can be directly obtained. For example, the coordinates of the non-contact temperature sensor relative to the center of the battery are (x, y, z), which can be recorded as a value xyz. Assuming the relative coordinates are (0, 1, 2) recorded as 012=12, the relative coordinates are (1, 0, 3) recorded as 103, and the relative coordinates are (1, 2, 0) recorded as 120.
[0054] Furthermore, the computing system also includes an ambient temperature sensor for obtaining ambient temperature and / or an ambient humidity sensor for obtaining ambient humidity and / or a light intensity sensor for obtaining light intensity and / or a wind speed sensor for obtaining wind speed.
[0055] In addition, the computing system also includes a cloud module; the cloud module can receive information obtained by the environmental data acquisition module and the battery data acquisition module, and transmit the received information to the processor module. After the target battery is stored, the target battery material, battery color, battery cleanliness (by manual judgment) and other battery data can also be manually input, and then these data are transmitted to the cloud. When calculating the surface temperature of the target battery, the target battery data can be directly obtained from the cloud.
[0056] Furthermore, the non-contact temperature sensor needs to meet the following requirements: it can work stably in a working environment of -40 to 60 degrees Celsius, and can accurately collect the surface temperature of objects between -50 and 300 degrees Celsius; and it has a fixed relative position with the battery under test, and the detection field of the non-contact temperature sensor can cover the battery area.
[0057] In addition, the computing system also includes a local storage module, which can interact with the processor module and the cloud module.
[0058] In a third aspect, the present invention further claims protection for a battery storage device, the battery storage device comprising the battery surface temperature calculation system of any one of the above embodiments, and capable of executing the battery surface temperature calculation method of any one of the above embodiments.
[0059] The battery storage device can be a battery swap station with battery storage function, or a box that only stores batteries, or a special vehicle that can transport batteries (including a box for storing batteries), or other types of equipment that can store batteries.
[0060] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.
Claims
1. A method for calculating the surface temperature of a battery, characterized in that: The calculation method includes: Acquire multiple sets of training temperatures and multiple sets of historical characteristic data on the battery surface, wherein the training temperatures correspond to the historical characteristic data one by one, and each set of historical characteristic data includes environmental data and battery data; Performing neural network training based on the training temperature and historical feature data corresponding to the training temperature to obtain a temperature calculation model; Obtain real-time environmental characteristic data of the space where the battery is located and real-time characteristic data of the battery; The real-time characteristic data is input into the temperature calculation model to calculate and obtain the real-time battery surface temperature.
2. The calculation method according to claim 1, characterized in that: The environmental data includes environmental temperature and / or environmental humidity and / or light intensity and / or wind speed.
3. The calculation method according to claim 1, characterized in that: The battery data includes battery material and / or battery color and / or battery storage location and / or battery cleanliness and / or measured battery surface temperature.
4. The calculation method according to claim 3, characterized in that: The temperature calculation model includes multiple hidden layers, each hidden layer includes multiple neurons, and the input of each neuron is Output is H k,i =φ(G k,i ), After K hidden layers, the output is y; Among them, G k,i represents the input of the i-th neuron in the K-th hidden layer, is the weight of the jth input data in the ith neuron in the kth hidden layer, x j is the jth input data, represents the input bias term of the i-th neuron in the k-th layer; n k-1 Indicates the number of input data in the kth layer.
5. A battery surface temperature calculation system, characterized in that: The computing system comprises: Environmental data acquisition module: capable of acquiring environmental data; Battery data acquisition module: capable of acquiring battery data; Processor module: Based on the acquired environmental data and battery data, the real-time battery surface temperature is calculated using a temperature calculation model.
6. The computing system according to claim 5, characterized in that: The battery data acquisition module includes a non-contact temperature sensor, which can directly measure the battery surface temperature before correction; The processor module calculates the corrected real-time battery surface temperature using a temperature calculation model based on the battery surface temperature before correction and environmental data.
7. The computing system according to claim 5, characterized in that: The battery data acquisition module includes a scanning device, and an identification image is provided on the battery. The scanning device can obtain the battery material and / or battery color after scanning the identification image.
8. The computing system according to claim 5, characterized in that: The computing system also includes a cloud module; The cloud module can receive the information acquired by the environment data acquisition module and the battery data acquisition module, and transmit the received information to the processor module.
9. The computing system according to claim 5, characterized in that: The environmental data acquisition module includes: an environmental temperature sensor and / or an environmental humidity sensor and / or a light intensity sensor and / or a wind speed sensor.
10. A battery storage device, characterized in that: The battery storage device comprises the battery surface temperature calculation system according to any one of claims 5 to 9, and is capable of executing the battery surface temperature calculation method according to any one of claims 1 to 4.