Safety monitoring method for lithium battery in container

Abstract

The invention belongs to the technical field of smart freight, and discloses a safety monitoring method for a lithium battery in a container, which comprises the following steps: acquiring environmental data in the container, battery external state data and battery internal data, the environmental data comprising in-container environmental temperature data and in-container environmental pressure data, the battery state data comprises battery surface temperature data, battery position data and battery appearance characteristic data; performing data fusion on the environment data and the external state data of the battery to obtain external fusion data of the battery; and constructing a battery state prediction model based on the battery external fusion data and the battery internal data, and executing real-time prediction of the internal and external states of the battery based on the constructed battery state prediction model. According to the invention, continuous monitoring of the environment where the power lithium battery is located is realized, and the monitoring strategy can be dynamically adjusted to adapt to continuously changing environmental conditions. Therefore, comprehensive risk assessment and management are carried out.

Description

Technical Field

[0001] The present invention belongs to the field of smart freight technology, and in particular relates to a safety monitoring method for lithium batteries in a container. Background Art

[0002] In the cross-regional railway transportation scenario, special conditions such as severe temperatures, external impacts or collisions, and extreme environmental pressures pose a severe challenge to the safety of power lithium batteries. Given the intertwined influence of these complex environmental factors, traditional video surveillance technology is difficult to accurately monitor the safety status of power lithium batteries. Therefore, there is an urgent need for a method that can efficiently and real-time evaluate the status of power lithium batteries. Summary of the invention

[0003] The purpose of the present invention is to provide a safety monitoring method for lithium batteries in a container to solve the problems existing in the above-mentioned prior art.

[0004] To achieve the above-mentioned purpose, the present invention provides a safety monitoring method for lithium batteries in a container, comprising: obtaining environmental data in the container, battery external status data and battery internal data, the environmental data including internal environment temperature data and internal environment pressure data, and the battery status data including battery surface temperature data, battery position data and battery appearance feature data; performing data fusion on the environmental data and the battery external status data to obtain battery external fusion data; constructing a battery status prediction model based on the battery external fusion data and the battery internal data, and performing real-time prediction of the internal and external states of the battery based on the constructed battery status prediction model.

[0005] Optionally, the data fusion of the environmental data and the battery external state data specifically includes: preprocessing the in-box environmental temperature data and the in-box environmental pressure data respectively to obtain preprocessed environmental data;

[0006] Establishing a data model of the battery surface temperature data, the battery position data and the battery appearance characteristic data;

[0007] The preprocessed environmental data and the established data model are fused and registered respectively to obtain the registered environmental data and the registered battery external state data, and the registered data are fused to obtain the battery external fused data.

[0008] Optionally, the preprocessing process of the in-box ambient temperature data specifically includes: grayscale processing the in-box ambient temperature data to obtain grayscale image data, establishing a conversion relationship between grayscale value and temperature based on the grayscale image data, and obtaining preprocessed in-box ambient temperature data.

[0009] Optionally, the preprocessing process of the in-box environmental pressure data specifically includes: performing signal amplification processing, normalization processing and analog-to-digital conversion processing on the in-box environmental pressure data in sequence to obtain preprocessed in-box environmental pressure data.

[0010] Optionally, the method for constructing a data model corresponding to the battery surface temperature data specifically includes:

[0011] The battery surface is divided into upper, lower, left, right and center regions, and the temperature state vector is obtained based on the temperature and temperature change rate of each region:

[0012]

[0013] Where, T r (k) is the temperature of the rth region at time step k, is the temperature change rate of the rth region at time step k; T k is the state vector at time k;

[0014] A state transition model is established based on the temperature state vector:

[0015] No control input:

[0016] With control input:

[0017] In the formula, is the control input vector at time k, is the process noise, F is the state transfer matrix;

[0018] Establish the data model corresponding to the battery surface temperature data:

[0019]

[0020] In the formula, is the observation vector at time step k, is the observation noise, and H is the observation matrix.

[0021] Optionally, the method for constructing the data model corresponding to the battery location data specifically includes:

[0022] Based on the Kalman filter combined with the displacement sensor to determine the position of the power lithium battery,

[0023]

[0024] In the formula, S k is the state vector, [x k ,y k ] is the position coordinate of the center of the power lithium battery;

[0025] Define the state transition model:

[0026] No control input:

[0027] With control input:

[0028] In the formula, S k+1 is the state vector at time step k+1, F is the state transfer matrix, is the process noise; B is the control matrix, is the control input vector;

[0029] Define the observation model:

[0030]

[0031] In the formula, is the measured value of the displacement sensor, is the observation noise, and H is the observation matrix.

[0032] Optionally, the method for constructing the data model corresponding to the battery appearance feature data specifically includes:

[0033] Grayscale processing and contrast adjustment of battery appearance feature data:

[0034] I gray =0.2989×R+0.5870×G+0.1140×B

[0035] I new =a·(I old -I min )+I max

[0036] In the formula, I gray is the converted grayscale value, R, G and B are the intensity values ​​of the red, green and blue channels of the original RGB image at the corresponding pixel points respectively; I new is the enhanced image intensity, a is the stretching ratio, I old is the intensity value of the original image, I min is the minimum intensity value of the original image, I max is the maximum intensity value of the original image;

[0037] Get the appearance features of the power lithium battery and construct the state vector:

[0038]

[0039] In the formula, B k is the state vector, W k is the battery width, H kis the battery height, θ k For battery direction;

[0040] Define the state transition model:

[0041] Still image: B k+1 =F·B k +w Bk

[0042] Dynamic image: B k+1 =F·B k +B·u Bk +w Bk

[0043] In the formula, B k+1 is the state vector at time step k+1, F is the state transfer matrix, w Bk is the process noise, B is the control input matrix, u Bk is the control input vector;

[0044] Define the observation model:

[0045] Z Bk =H·B k +v Bk

[0046] In the formula, Z Bk is the observation vector, H is the observation matrix, v Bk is the observation noise.

[0047] Optionally, the process of acquiring the registered environmental data specifically includes:

[0048] The preprocessed environmental data is weighted averaged to obtain the registered environmental data.

[0049] Optionally, the process of fusing and registering the established data model specifically includes:

[0050] Constructing a state transition model and an observation model based on the state vector corresponding to the battery surface temperature data, the battery position data and the battery appearance feature data;

[0051] Determine the initial state estimate and the initial estimate error covariance, predict the state of the next time step based on the state transition model, and predict the current state estimate and the current prediction error covariance;

[0052] Calculating a Kalman gain based on real-time battery external state data, and determining a contribution degree of the real-time battery external state data to the predicted state based on the Kalman gain;

[0053] Calculate the observation difference between the predicted observation value of the observation model and the actual observation value, and update the initial state estimate and the initial estimation error covariance based on the observation difference;

[0054] The prediction and update process is repeated until the observation data of the battery's surface temperature, position, and appearance characteristics are integrated into the state estimation, completing the data fusion registration.

[0055] Optionally, building a battery state prediction model based on the battery external fusion data and the battery internal data specifically includes:

[0056] Performing standardization processing on the battery external fusion data and the battery internal data to obtain preprocessed data;

[0057] A battery internal indicator vector is constructed based on the preprocessed data, and a state transfer equation and an observation model are defined based on the constructed battery internal indicator vector;

[0058] Based on the state transfer equation and the observation model, the posterior probability of the battery state is calculated in combination with the Bayesian update model, the Bayesian update model is updated based on the posterior probability, and the real-time prediction of the internal and external states of the battery is performed according to the updated Bayesian update model. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0060] Figure 1 This is an overall flow chart of railway container power lithium battery safety monitoring based on multi-source data fusion in an embodiment of the present invention;

[0061] Figure 2 This is a deployment scheme for the data capture device on a railway container in an embodiment of the present invention;

[0062] Figure 3 A schematic diagram of the relationship between various types of data preprocessing collected in an embodiment of the present invention;

[0063] Figure 4 is a data registration flow chart in an embodiment of the present invention, wherein: Figure 4 (a) is a schematic diagram of a multi-source data registration algorithm for the internal ambient temperature and pressure of a container in an embodiment of the present invention; Figure 4 (b) is a schematic diagram of a multi-source data registration algorithm for battery surface temperature, position and appearance feature data in an embodiment of the present invention;

[0064] Figure 5 A schematic diagram of multi-source data combined and fused by an ultrasonic sensor in an embodiment of the present invention;

[0065] Figure 6 This is a schematic diagram of judging the internal and external conditions of a power lithium battery in an embodiment of the present invention;

[0066] Figure 7 It is a flowchart of implementing safety monitoring in an embodiment of the present invention. DETAILED DESCRIPTION

[0067] Various exemplary embodiments of the present invention will now be described in detail. This detailed description should not be considered as limiting the present invention, but should be understood as a more detailed description of certain aspects, features, and embodiments of the present invention.

[0068] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0069] Embodiment 1

[0070] like Figure 1 - Figure 7 As shown, this embodiment provides a safety monitoring method for lithium batteries in a container, including: obtaining environmental data in the container, battery external status data and battery internal data, the environmental data including box environment temperature data and box environment pressure data, the battery status data including battery surface temperature data, battery position data and battery appearance feature data; performing data fusion on the environmental data and the battery external status data to obtain battery external fusion data; constructing a battery status prediction model based on the battery external fusion data and the battery internal data, and performing real-time prediction of the internal and external states of the battery based on the constructed battery status prediction model.

[0071] This embodiment proposes a monitoring method that combines the ambient temperature data and ambient pressure data in the railway container with the surface temperature data, location data and appearance feature data of the power lithium battery. This method deeply integrates multiple key data sources and aims to achieve efficient and real-time evaluation of the safety of power lithium batteries.

[0072] The purposes of this embodiment are: (1) to provide a practical application solution to fill the application gap of power lithium batteries in the field of railway transportation; (2) to use multi-source data fusion technology to solve the storage and transportation management problems of power lithium batteries; (3) to realize the Internet of Things monitoring of railway containers to ensure timely response to transportation anomalies.

[0073] This embodiment monitors the railway container environment and battery status based on multiple devices. Infrared thermal imagers detect the environment inside the box and the surface temperature of the battery; pressure sensors measure the environmental pressure; cameras observe the outside of the battery; and ultrasonic sensors detect the inside of the battery. Data is wirelessly transmitted to the central processing system, and multi-source data fusion improves data accuracy and response speed to ensure railway transportation safety. The specific technical solutions are as follows:

[0074] The overall process of this method is as follows Figure 1 As shown. Data capture equipment is deployed to collect and preprocess data such as ambient temperature and pressure in railway containers. Data of different physical quantities are fused, and the fused data and algorithms are used to predict battery conditions. The prediction accuracy is optimized through weighted average, Kalman filter and Bayesian method, and the identification results are transmitted to the central processing system via 5G or Bluetooth to facilitate supervision and grasp of battery conditions. The steps are as follows:

[0075] Deployment of data capture equipment: The deployment plan of the equipment is as follows Figure 2 As shown. An infrared thermal imager, a high-definition camera, an ultrasonic sensor, and a temperature and pressure regulating switch are installed in the container; a displacement sensor and a position regulating switch are installed under the power lithium battery; and a pressure sensor and a central processing system are arranged nearby.

[0076] S100, press Figure 2 The equipment shown in the arrangement obtains the ambient temperature, pressure, and battery position data collected by the infrared thermal imager, pressure sensor, displacement sensor, etc., as well as the battery appearance data collected by the high-definition camera.

[0077] S200, pre-processing the collected railway container internal environment data, specifically including: Figure 3 The figure shows the relationship between the preprocessing of various types of collected data.

[0078] S201, preprocessing of the ambient temperature data collected by the infrared thermal imager in the box. The specific steps are as follows:

[0079] (1) Image grayscale: Convert the thermal imaging image of the ambient temperature inside the box into a grayscale image to simplify the data and reduce processing complexity, thereby improving analysis efficiency.

[0080] G=0.299*M r +0.587*M g +0.114*M b

[0081] Where G is the calculated grayscale value, M r 、M g 、M b are the pixel values ​​of the red, green, and blue channels in the original image, respectively.

[0082] (2) Establish the conversion relationship between gray value and temperature:

[0083]

[0084] T 内 is the internal temperature of the container, G is the gray value of the image, T max is the maximum temperature inside the container, and the minimum temperature is T min , G max and G min They correspond to the maximum grayscale value and the minimum grayscale value of the image respectively.

[0085] S202: Preprocessing of the ambient pressure data collected by the pressure sensor. The specific steps are as follows:

[0086] (1) Signal amplification: Enhance the output signal and amplify the signal to facilitate pressure detection in the box:

[0087] V out =A·V in

[0088] V out is the amplified output voltage, A is the amplification factor, V in is the input voltage.

[0089] (2) Environmental pressure normalization and data conversion: helps to adjust the data into a format that is easy to analyze.

[0090] ① Normalization processing: Scale the data to a specific range to improve the convergence speed and accuracy of the algorithm.

[0091]

[0092] Where P norm is the normalized environmental pressure data, P 原 is the original ambient pressure data, P max and P min They are the maximum and minimum values ​​of the pressure data respectively.

[0093] ②Conversion of analog signals to digital signals: Convert analog output signals into digital signals for easy processing.

[0094] D=ADC(V in )

[0095] Where D is the digital value, ADC value is the reading of the ADC register, and V in is the input voltage.

[0096]

[0097] Where P is the pressure value, k is the proportional coefficient of the pressure sensor, Vmax and V min are the maximum and minimum values ​​of the ADC input voltage respectively, and b is the voltage output of the sensor at zero pressure.

[0098] S300 pre-processes the collected power lithium battery data, including:

[0099] (1) In the Kalman filter, both process noise and observation noise are Gaussian white noise. The formula is:

[0100] ①Process noise vector w k : Uncertainty in the state prediction at time step k.

[0101]

[0102] In the formula, w Tk is the process noise of the battery surface temperature, is the process noise at the battery location, It is the process noise that is characteristic of the battery appearance.

[0103] ②Observation noise vector v k : The uncertainty in the observation at time step k.

[0104]

[0105] In the formula, is the observation noise of the battery surface temperature, is the observation noise of the battery position, is the observation noise of the battery appearance characteristics.

[0106] (2) Controlling the input vector u in using the Kalman filter k , control matrix B and observation matrix H are expressed as:

[0107] ①Control input vector u k And the control matrix B:

[0108]

[0109] In the formula, are the control inputs that affect the battery surface temperature, position, and appearance characteristics. The non-zero elements on the diagonal represent and Affects battery temperature, position and appearance characteristics.

[0110] ②Observation matrix H:

[0111]

[0112] The non-zero elements on the diagonal represent the characteristic states that affect the battery temperature, position, and appearance.

[0113] (3) The state transfer matrix F describes the evolution of the state variables in a small time interval Δt. It is expressed as follows:

[0114]

[0115] Δt reflects the effect of the temperature change rate on the temperature. The 1 on the diagonal indicates the continuity of the state itself, and 0 indicates no effect on the state transition.

[0116] S301, establish a battery surface temperature data model, the specific steps are as follows:

[0117] (1) Image acquisition: The infrared thermal imager captures clear thermal images of the battery, covering the entire surface.

[0118] (2) Image preprocessing

[0119] ①Contrast adjustment:

[0120] By adjusting parameters, the image contrast is optimized, hot and cold spots are highlighted, and the temperature distribution is analyzed.

[0121] I′=a·(II min )+b

[0122] Where I′ is the new image after contrast stretching, a is the proportional constant, I is the original image, and I min is the minimum intensity value of the original image, and b is the offset value, which adjusts the overall brightness of the stretched image.

[0123] ② Edge detection: The Sobel operator helps identify hot spots and abnormal temperature areas on the battery surface by detecting changes in temperature gradients.

[0124]

[0125] Where I is the original image, G x *I and G y *I are the results of the horizontal and vertical convolution of the image with Sobel operators.

[0126] (3) Define the state vector: Obtain the surface temperature data of the power lithium battery from the processed image.

[0127] The battery surface is divided into five areas: top, bottom, left, right and center areas, which are expressed as:

[0128]

[0129] T r (k) is the temperature of the rth region at time step k, is the rate of temperature change of the rth region at time step k.

[0130] (4) Establishing a state transition model: In the Kalman filter, the state transition model of the infrared thermal imager describes the evolution of the temperature state over time. The following is a basic mathematical representation:

[0131] (No control input)

[0132] (with control input)

[0133] T k+1 is the state vector at time step k+1, is the process noise, is the control input vector.

[0134] (5) Establish observation model: Infrared thermal imager provides the measurement value of battery surface temperature

[0135] In the formula, is the observation vector at time step k, is the observation noise.

[0136] S302 establishes a battery location data model. The specific steps are as follows:

[0137] (1) Define the state vector:

[0138] State vector:

[0139] In the formula, [x k ,y k ] are the position coordinates of the center of the power lithium battery.

[0140] (2) Define the state transition model:

[0141] (No control input)

[0142] (with control input)

[0143] S k+1 is the state vector at time step k+1, is the control input vector, is the process noise.

[0144] (3) Define the observation model: the measurement value of the displacement sensor With the state vector S k The observation model is as follows:

[0145]

[0146] is the observation vector at time step k, S k is the state vector, is the observation noise.

[0147] S303, establishing a battery appearance feature data model, the specific steps are as follows:

[0148] (1) Image acquisition: A high-definition camera captures images of battery appearance features.

[0149] (2) Image grayscale: improve contrast and clarity, simplify processing and reduce calculation amount.

[0150] I gray =0.2989×R+0.5870×G+0.1140×B

[0151] I gray is the converted grayscale value, R, G, and B are the intensity values ​​of the red, green, and blue channels of the original RGB image at the corresponding pixel points, respectively.

[0152] (3) Contrast adjustment: Enhance the image contrast to make the battery shape more obvious and the light and dark distinct.

[0153] I new =a·(I old -I min )+I max

[0154] I new is the enhanced image intensity, a is the stretching ratio, I old ,I min and I max are the intensity value, minimum intensity value, and maximum intensity value of the original image respectively.

[0155] (4) Define the state vector: obtain the appearance features of the power lithium battery from the processed image.

[0156] Let the state vector B k Width W including power lithium battery k , height H k and direction θ k .

[0157] State vector:

[0158] (5) Define the state transition model:

[0159] (Still image)

[0160] (Motion Picture)

[0161] B k+1 is the state vector at time step k+1, is the control input vector, is the process noise.

[0162] (6) Define the observation model: observation vector Contains the battery appearance feature data. The observation model is:

[0163]

[0164] In the formula, is the observation vector at time step k, is the observation noise.

[0165] S400, perform multi-source data fusion registration algorithm on the collected data. The specific steps are as follows:

[0166] Various types of data are used for multi-source data registration algorithms such as Figure 4 (a)(b) Schematic diagram.

[0167] S401, multi-source data fusion registration algorithm for the internal environment temperature and pressure of the container.

[0168] (1) Determine weights: assign different weights to temperature and pressure, and calculate the weighted average to obtain fused data.

[0169]

[0170] and W P are the weights of ambient temperature data and ambient pressure data respectively.

[0171] (2) Calculate the weighted average:

[0172] In the formula, FR is the fusion result, T 内 and P are the ambient temperature and pressure data inside the box respectively.

[0173] (3) Verify and adjust the fused data: Analyze the weighted average data to ensure the accuracy of the fused data.

[0174] S402, multi-source data fusion registration algorithm for battery surface temperature, position and appearance feature data.

[0175] (1) Define the multivariable state vector:

[0176] X k =[T k S k B k ]T

[0177] Where, T k is the battery surface temperature and its rate of change, S k Represents the position coordinates of the center of the power lithium battery, B k Represents the appearance characteristics of power lithium batteries.

[0178] (2) Establish state transfer model and observation model:

[0179] ① State transition model: The battery surface temperature, position and appearance characteristics can be regarded as a dynamic system:

[0180] X k+1 =F·X k +w k (No control input)

[0181] X k+1 =F·X k +B·u k +w k (with control input)

[0182] X k is the state vector at time k, u k is the control input vector at time k, w k is the process noise.

[0183] ②Observation model:

[0184] Define a multivariate observation vector:

[0185] Represent the observed values ​​of the surface temperature, position and appearance characteristics of the battery respectively.

[0186] For the surface temperature, position and appearance characteristics of the battery, the measurement equation of the observation model is expressed as:

[0187] Z k ′=H·X k +v k

[0188] Z k ′ is the observation vector, H is the observation matrix, v k is the observation noise.

[0189] (3) Initialize the Kalman filter: Set the initial state estimate and the estimated error covariance matrix P 0 .

[0190] (4) Kalman filter prediction step: first use the dynamic model to predict the state of the next time step.

[0191] ① Predict the current state estimation: Use the state transition model to predict the state at the next moment

[0192] (No control input)

[0193] (with control input)

[0194] In the formula, is the posterior state estimate at the previous moment.

[0195] ② Forecast error covariance: predict the error covariance at the current moment, taking into account the impact of process noise

[0196] P K|K-1 =F·P K-1|K-1 ·F T +Q

[0197] Where P K|K-1 is the forecast error covariance, and Q is the process noise covariance matrix.

[0198] (5) Kalman filter update steps: Observation data collection: Collect observation data Z k ′.

[0199] ① Calculate the Kalman gain: the weight that determines how to fuse the observed data into the predicted state.

[0200] K k =P K|K-1 H T ·(HP K|K-1 H T +R) -1

[0201] In the formula, K k is the Kalman gain, and R is the observation noise covariance matrix.

[0202] ②Calculate the observed residual value:

[0203] Where Y k is the observed residual value, Z k ′ is the observation vector at the current moment.

[0204] ③Update state estimation:

[0205] In the formula, is the predicted state, is the updated state estimate.

[0206] ④ Update the estimated error covariance: reflect the uncertainty after integrating the new observation data:

[0207] P K|K =(IK k H)P K|K-1

[0208] Where P K|K is the posterior error covariance, and I is the identity matrix

[0209] (6) Iterative process: Repeat the prediction and update steps to achieve continuous estimation and fusion of the system state.

[0210] (7) Data fusion: Through the prediction and update steps, the observation data such as battery surface temperature and position are integrated into the state estimation. The Kalman gain determines the contribution of the observation data to the final estimate, achieving optimal fusion.

[0211] (8) Result analysis and decision-making: By analyzing the updated state estimate and error covariance, the state of the power lithium battery in the railway container is predicted, the parameters are adjusted, and the optimal state estimate is provided.

[0212] S500, ultrasonic sensors combine with fusion data to predict the internal condition of the power lithium battery.

[0213] like Figure 5 Shown is a schematic diagram of ultrasonic sensor combined with fusion data.

[0214] This embodiment proposes that the ultrasonic sensor is combined with fusion data to predict the internal condition of the battery in the following steps:

[0215] (1) Data collection: The ultrasonic sensor collects the original internal data of the battery and the fused data from the above step S400.

[0216] (2) Data processing: The collected fusion data is standardized to ensure quality and consistency.

[0217]

[0218] In the formula, and FR are the normalized data,

[0219] norm max(FR) and min(FR) is the maximum and minimum value of each data.

[0220] The raw internal data of the power lithium battery collected by the ultrasonic sensor is processed:

[0221] ①Define state variables: The state vector x(k) includes various internal indicators of the battery

[0222]

[0223] ②Define the state transfer equation: Assume that the state change of the battery follows certain physical laws and is expressed as:

[0224]

[0225] In the formula, is the time derivative of the state vector, x(k) is the battery state vector, A is the state transfer matrix, B is the control input matrix, u(k) is the control input vector, and w(k) is the process noise vector.

[0226]

[0227] In the formula, δ 1 represents the state change rate, b nm represents the influence of the control input u(k) on the state variable x(k), u m (k) is the mth control input at time k, Q is the covariance matrix of the process noise, σ wk 2 is the variance of the kth process noise random variable.

[0228] ③Define the observation model: The observation model links the device measurements to the battery status:

[0229] z(k)=H·x(k)+v(k)

[0230] Where z(k) is the observation vector, H is the observation matrix, and v(k) is the observation noise.

[0231]

[0232] In the formula, h mn is the mapping weight from the state variable x(k) to the observation vector z(k).

[0233] (3) Establishing Bayesian updating model:

[0234] Status update process:

[0235]

[0236] In the formula, is the posterior probability, is the joint likelihood, p(x(k)) is the prior probability, is the marginal probability.

[0237] (4) Prediction and health management: Use the updated model to predict the status of the power lithium battery in real time, and make health management and maintenance decisions based on it.

[0238] S600, recognition result feedback

[0239] like Figure 6 The figure shows a schematic diagram for judging the internal and external conditions of a power lithium battery.

[0240] The recognition results and multi-source data obtained in step S500 are fed back to the supervisors through the central processing system and transmitted wirelessly to form a data set, providing data reference for safety monitoring of batteries in railway container transportation.

[0241] This method integrates the data of equipment such as infrared thermal imagers and pressure sensors, and combines weighted average, Kalman filter and Bayesian method to construct a power lithium battery railway container transportation safety monitoring method based on multi-source data fusion technology. This method can monitor and feedback the battery status and environmental changes in the container in real time. Equipment selection and application: Select diversified sensors (pressure, displacement, ultrasonic sensors) and infrared thermal imagers, supplemented by high-definition cameras and temperature, pressure and position adjustment switches. Data transmission method: Use 5G, Bluetooth and other transmission methods to ensure the speed and timeliness of data; at the data processing level: the central processing system integrates data and determines the battery status.

[0242] The above is only a preferred specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A safety monitoring method for lithium batteries in containers, characterized in that: include: Obtaining environmental data in the container, battery external status data and battery internal data, wherein the environmental data includes internal environment temperature data and internal environment pressure data, and the battery status data includes battery surface temperature data, battery position data and battery appearance feature data; Performing data fusion on the environmental data and the battery external state data to obtain battery external fusion data; A battery state prediction model is constructed based on the battery external fusion data and the battery internal data, and real-time prediction of the internal and external states of the battery is performed based on the constructed battery state prediction model.

2. A safety monitoring method for lithium batteries in a container according to claim 1, characterized in that: The data fusion of the environmental data and the battery external state data specifically includes: Preprocessing the in-box ambient temperature data and the in-box ambient pressure data respectively to obtain preprocessed ambient data; Establishing a data model of the battery surface temperature data, the battery position data and the battery appearance characteristic data; The preprocessed environmental data and the established data model are fused and registered respectively to obtain the registered environmental data and the registered battery external state data, and the registered data are fused to obtain the battery external fused data.

3. A safety monitoring method for lithium batteries in a container according to claim 2, characterized in that: The preprocessing process of the ambient temperature data in the box specifically includes: The box interior environment temperature data is gray-scale processed to obtain gray-scale image data, and a conversion relationship between gray-scale value and temperature is established based on the gray-scale image data to obtain pre-processed box interior environment temperature data.

4. A safety monitoring method for lithium batteries in a container according to claim 2, characterized in that: The preprocessing process of the in-box environmental pressure data specifically includes: The in-box environmental pressure data is subjected to signal amplification processing, normalization processing and analog-to-digital conversion processing in sequence to obtain preprocessed in-box environmental pressure data.

5. A safety monitoring method for lithium batteries in a container according to claim 2, characterized in that: The method for constructing the data model corresponding to the battery surface temperature data specifically includes: The battery surface is divided into upper, lower, left, right and center regions, and the temperature state vector is obtained based on the temperature and temperature change rate of each region: Where, T r (k) is the temperature of the rth region at time step k, is the temperature change rate of the rth region at time step k; T k is the state vector at time k; A state transition model is established based on the temperature state vector: No control input: With control input: In the formula, is the control input vector at time k, is the process noise, F is the state transfer matrix; Establish the data model corresponding to the battery surface temperature data: In the formula, is the observation vector at time step k, is the observation noise, and H is the observation matrix.

6. A safety monitoring method for lithium batteries in a container according to claim 2, characterized in that: The method for constructing the data model corresponding to the battery location data specifically includes: Based on the Kalman filter combined with the displacement sensor to determine the position of the power lithium battery, In the formula, S k is the state vector, [x k ,y k ] is the position coordinate of the center of the power lithium battery; Define the state transition model: No control input: With control input: In the formula, S k+1 is the state vector at time step k+1, F is the state transfer matrix, is the process noise; B is the control matrix, is the control input vector; Define the observation model: In the formula, is the measured value of the displacement sensor, is the observation noise, and H is the observation matrix.

7. A safety monitoring method for lithium batteries in a container according to claim 2, characterized in that: The method for constructing the data model corresponding to the battery appearance feature data specifically includes: Grayscale processing and contrast adjustment of battery appearance feature data: I gray =0.2989×R+0.5870×G+0.1140×B I new =a·(I old -I min )+I max In the formula, I gray is the converted grayscale value, R, G and B are the intensity values ​​of the red, green and blue channels of the original RGB image at the corresponding pixel points respectively; I new is the enhanced image intensity, a is the stretching ratio, I old is the intensity value of the original image, I min is the minimum intensity value of the original image, I max is the maximum intensity value of the original image; Get the appearance features of the power lithium battery and construct the state vector: In the formula, B k is the state vector, W k is the battery width, H k is the battery height, θ k For battery direction; Define the state transition model: Still Image: Dynamic images: In the formula, B k+1 is the state vector at time step k+1, F is the state transfer matrix, is the process noise, B is the control input matrix, u Bk is the control input vector; Define the observation model: Z Bk =H·B k +v Bk In the formula, Z Bk is the observation vector, H is the observation matrix, v Bk is the observation noise.

8. A safety monitoring method for lithium batteries in a container according to claim 2, characterized in that: The process of acquiring the registered environmental data specifically includes: The preprocessed environmental data is weighted averaged to obtain the registered environmental data.

9. A safety monitoring method for lithium batteries in a container according to claim 2, characterized in that: The process of fusing and registering the established data model specifically includes: Constructing a state transition model and an observation model based on the state vector corresponding to the battery surface temperature data, the battery position data and the battery appearance feature data; Determine the initial state estimate and the initial estimate error covariance, predict the state of the next time step based on the state transition model, and predict the current state estimate and the current prediction error covariance; Calculating a Kalman gain based on real-time battery external state data, and determining a contribution degree of the real-time battery external state data to the predicted state based on the Kalman gain; Calculate the observation difference between the predicted observation value of the observation model and the actual observation value, and update the initial state estimate and the initial estimation error covariance based on the observation difference; The prediction and update process is repeated until the observation data of the battery's surface temperature, position, and appearance characteristics are integrated into the state estimation, completing the data fusion registration.

10. A safety monitoring method for lithium batteries in a container according to claim 1, characterized in that: Building a battery state prediction model based on the battery external fusion data and the battery internal data specifically includes: Standardizing the battery external fusion data and the battery internal data to obtain preprocessed data; A battery internal indicator vector is constructed based on the preprocessed data, and a state transfer equation and an observation model are defined based on the constructed battery internal indicator vector; Based on the state transfer equation and the observation model, the posterior probability of the battery state is calculated in combination with the Bayesian update model, the Bayesian update model is updated based on the posterior probability, and the real-time prediction of the internal and external states of the battery is performed according to the updated Bayesian update model.

Images