A temperature monitoring method and system for a new energy load cable

Abstract

The present application relates to the field of cable temperature monitoring, more particularly, the present application relates to a new energy load cable temperature monitoring method and system. The method comprises: obtaining a temperature sequence of a preset period; calculating a first-order difference sequence of the history sequence, iteratively segmenting the first-order difference sequence, constructing a segmentation evaluation function, obtaining a plurality of sub-difference sequences, and constructing a network model for any sub-difference sequence; inputting the to-be-predicted sequence into any network model to output the corresponding temperature prediction value, taking the importance of the calculated network model as the weight, calculating the product of the weight and the temperature prediction value, taking the sum of the products of each network model as the final temperature prediction value, and completing the temperature monitoring. Through the technical scheme of the present application, the accuracy of temperature prediction can be improved, and the response speed and stability of temperature monitoring can be improved.

Description

Technical Field

[0001] This invention relates to the field of cable temperature monitoring. More specifically, this invention relates to a method and system for monitoring the temperature of load cables used in new energy applications. Background Technology

[0002] Charging cables play a crucial role as a key component connecting electric vehicles and charging stations. During electric vehicle charging, the charging cable, as the carrier of electrical energy, faces the risk of overheating. As current flows through the cable, it generates heat, especially during frequent high-current flows in the charging process, which can cause the internal temperature of the cable to rise sharply. This temperature increase can not only affect the long-term service life of the cable but also lead to cable damage or even serious safety accidents such as fires. Abnormal temperature increases can cause insulation aging, short circuits, and fires, directly threatening the lives and property of users.

[0003] A Chinese patent application with publication number CN115455797A discloses a method, apparatus, and electronic device for temperature prediction model training and temperature decision-making, including: inputting a candidate set temperature and acquired state parameters into a temperature prediction model, correcting the output data center temperature prediction value, and deciding on a second set temperature based on the corrected data center temperature prediction value, wherein the candidate set temperature is the temperature obtained by adjusting the first set temperature collected during temperature decision-making; when triggering model training, inputting the second set temperature and state parameters from the sample into the current temperature prediction model, and training the model with the data center temperature measurement value in the output sample as the target.

[0004] However, during the charging process of electric vehicles, temperature changes are usually affected by multiple factors and show different trends at different stages. Traditional temperature prediction models alone can only capture the overall temperature trend, while ignoring the details and complexity of temperature changes at each stage of the charging process, resulting in inaccurate temperature prediction results and thus affecting the accuracy of temperature monitoring results. Summary of the Invention

[0005] To address the problem of inaccurate temperature prediction results, this invention provides solutions in the following aspects.

[0006] In a first aspect, this invention discloses a method for temperature monitoring of load cables for new energy applications, comprising: acquiring a temperature sequence for a preset time period, the temperature sequence including a historical sequence and a sequence to be predicted; calculating a first-order difference sequence of the historical sequence, iteratively segmenting the first-order difference sequence, constructing a segmentation evaluation function, stopping the iteration when the segmentation evaluation function reaches its maximum, obtaining several sub-difference sequences, constructing a network model for any sub-difference sequence, with one sub-difference sequence corresponding to one network model; inputting the sequence to be predicted into any network model to output the corresponding temperature prediction value, using the importance of the calculated network model as a weight, calculating the product of the weight and the temperature prediction value, and using the sum of the products of each network model as the final temperature prediction value to complete temperature monitoring.

[0007] The first-order difference sequence of the historical data is calculated and iteratively segmented to extract multiple sub-difference sequences, each representing different temperature change patterns and trends. By constructing an independent network model for each sub-difference sequence, different dynamic characteristics of temperature changes can be captured, leading to more accurate temperature predictions. The final prediction result is obtained by calculating the importance of each network model and using it as a weight, multiplying it by the predicted temperature value, and summing the results. This weighted approach effectively integrates the advantages of multiple models, improving the overall prediction accuracy.

[0008] Preferably, the segmentation evaluation function satisfies the following relation:

[0009] , This represents the segmentation evaluation function. Sub-difference sequence The Middle Each difference value, Sub-difference sequence The mean of the differences, Sub-difference sequence The sequence length, This represents the total number of sub-difference sequences.

[0010] By designing a segmentation evaluation function, the partitioning of sub-difference sequences is effectively measured and optimized, ensuring that the internal differences within each sub-difference sequence are minimized, thereby improving the model's accuracy in capturing temperature changes. Maximizing the segmentation evaluation function achieves consistency among the sub-difference sequences, resulting in more stable and accurate temperature predictions.

[0011] Preferably, the segmentation evaluation function further includes: calculating the mean difference value of each sub-difference sequence, and calculating the variance of all mean difference values; the segmentation evaluation function satisfies the following relationship:

[0012] , This represents the segmentation evaluation function. Sub-difference sequence The Middle Each difference value, Sub-difference sequence The mean of the differences, Sub-difference sequence The sequence length, This represents the total number of sub-difference sequences. Represents an exponential function. This represents the variance of the mean of all differences.

[0013] By introducing the variance of the mean difference value and combining it with the exponential and normalization operations in the segmentation evaluation function, the partitioning of the sub-difference sequences is further optimized, ensuring the maximization of consistency within the sequences and the differences between sequences. Adding a variance term to the evaluation function effectively avoids over-segmentation or unreasonable segmentation, ensuring the stability of each sub-difference sequence in temperature change patterns, and enabling the prediction model to more accurately capture the patterns of temperature changes.

[0014] Preferably, the network model is a neural network or an ARIMA model.

[0015] Preferably, the importance includes: constructing a sliding window based on the length of the sequence to be predicted; dividing the historical sequence according to a preset sliding step size to obtain several windows; using any network model to predict each window to obtain the mean square error of each window; calculating the instability of each window; similarly obtaining the instability of the sequence to be predicted; clustering all windows and the sequence to be predicted based on the instability to obtain clusters; using the windows in the cluster where the sequence to be predicted is located as reference windows; and calculating the importance based on the mean square error of the reference windows.

[0016] Clustering all windows and the sequence to be predicted using instability creates different clusters. The sequence to be predicted can be assigned to a corresponding cluster based on its characteristics, allowing for the selection of a similar reference window. The mean squared error of these reference windows reflects their contribution to prediction accuracy and is thus used to calculate the importance of the network model.

[0017] Preferably, the importance satisfies the following relationship:

[0018] , Representation of network model The importance of Representation of network model In the reference window The mean square error, Indicates the total number of reference windows. This represents the total number of network models. This represents an exponential function.

[0019] Preferably, the instability includes: taking any sampling time as the target time, taking the previous sampling time adjacent to the target time as the adjacent time, taking the difference between the temperature at the target time and the temperature at the adjacent time as the temperature change rate at the target time, and obtaining a temperature change rate sequence for any window; calculating the cumulative difference between each temperature change rate and the mean temperature change rate in any window, and normalizing 0.01 and the sum of the cumulative difference values ​​as the window instability.

[0020] By analyzing the rate of temperature change at the target time and its adjacent times, subtle changes in temperature fluctuations can be captured, and potential irregularities or abrupt changes within the window can be revealed through instability assessment.

[0021] Preferably, it also includes: dividing the first-order difference sequence of the historical sequence according to a preset sliding step size to obtain several windows.

[0022] Preferably, the importance includes: using the first-order difference sequence of the sequence to be predicted as the temperature difference sequence, and calculating the distance between the temperature difference sequence and any sub-difference sequence; the importance satisfies the following relationship:

[0023] , Representation of network model The importance of Representing temperature difference sequences and sub-difference sequences distance, This represents the total number of sub-difference sequences. This represents an exponential function.

[0024] Secondly, the present invention discloses a temperature monitoring system for a load cable for new energy applications, comprising: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a temperature monitoring method for a load cable for new energy applications as described in any one of the present invention is implemented.

[0025] The beneficial effects of this invention are:

[0026] This invention iteratively segments and optimizes historical temperature data to ensure that the temperature fluctuation characteristics of different subsequences can be independently captured, thereby constructing the most suitable prediction model for each subsequence. This approach can effectively identify and respond to temperature change patterns over different time periods, enabling real-time temperature monitoring of cables during electric vehicle charging. Furthermore, by quantifying the importance of the network model, temperature prediction becomes not only more accurate but also allows for adaptive adjustment of monitoring strategies based on different temperature variations, further improving the safety and reliability of cable operation. Attached Figure Description

[0027] Figure 1 This is a flowchart of a method for monitoring the temperature of a load cable for new energy applications according to an embodiment of the present invention. Detailed Implementation

[0028] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0029] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0030] Reference Figure 1 A method for monitoring the temperature of load cables for new energy applications includes steps S1-S3, which are described in detail below.

[0031] S1: Obtain the temperature sequence for a preset time period, which includes historical sequences and sequences to be predicted.

[0032] It should be noted that the temperature rise of an electric vehicle is relatively gradual during the initial charging phase, but may fluctuate sharply in the later stages of charging or when the charging power is higher. These phased temperature changes often exhibit highly non-linear characteristics, and the rates of temperature change vary.

[0033] In one embodiment, during the charging process of an electric vehicle, a thermocouple sensor is used to collect the temperature of the cable, obtaining a temperature sequence over a preset time period. This temperature sequence includes historical temperature sequences and a temperature sequence to be predicted. The historical temperature sequence provides information on temperature changes over a past period, while the sequence to be predicted is the future temperature value to be predicted based on the current time and historical data.

[0034] Temperature data contains information on temperature changes at different stages of the production process. However, due to sensor measurement errors or environmental interference, the collected data often contains noise. To improve data quality, mean filtering technology is used to denoise the raw data, smoothing the data and removing abnormal fluctuations, thereby obtaining more accurate historical temperature data and the temperature to be predicted.

[0035] S2: Calculate the first-order difference sequence of the historical sequence, iteratively segment the first-order difference sequence, construct a segmentation evaluation function, stop iterating when the segmentation evaluation function is maximized, obtain several sub-difference sequences, and construct a network model for any sub-difference sequence.

[0036] It should be noted that, because temperature changes during charging exhibit different trends, traditional network models based on global historical temperature data are quite accurate in identifying temperature changes over long time spans, but their ability to identify local temperature fluctuations is weak, leading to biased prediction results. To more accurately predict temperature changes, the historical sequence is first differentially processed to extract the first-order difference sequence. Then, multiple random split points are used to divide the first-order difference sequence into multiple sub-difference sequences. By calculating the internal differences and inter-sequence differences of each sub-difference sequence, the optimal splitting result is evaluated and selected. Based on this splitting result, a separate network model is constructed for each sub-difference sequence.

[0037] In one embodiment, the segmentation evaluation function satisfies the following relation:

[0038] , This represents the segmentation evaluation function. Sub-difference sequence The Middle Each difference value, Sub-difference sequence The mean of the differences, Sub-difference sequence The sequence length, This represents the total number of sub-difference sequences.

[0039] By calculating the difference between each difference value in the same sub-difference sequence and the mean difference value of that sub-sequence, the intra-group variability within that sub-sequence can be effectively reflected. Specifically, a small intra-group variability means that the temperature change trend in the sub-difference sequence is relatively consistent, without obvious fluctuations or anomalous bands. This indicates that the internal variation pattern of the sub-difference sequence is relatively stable, the segmentation result is relatively accurate, and it can capture the characteristics of temperature change well. Conversely, a large intra-group variability indicates that the temperature fluctuation in the sub-difference sequence is relatively drastic or irregular, and there may be different trends or noise, resulting in an inaccurate segmentation result. Therefore, by minimizing the intra-group variability, we can ensure that each sub-difference sequence can better represent a relatively independent and consistent temperature change trend, thereby improving the accuracy of subsequent modeling and prediction.

[0040] To obtain the most accurate segmentation result by maximizing the segmentation evaluation function, several sub-segmentation sequences are obtained. A network model is constructed based on the temperature data corresponding to any sub-difference sequence. The construction, training, and optimization of the network model are existing technologies and will not be elaborated further here. It should be noted that the network model is a neural network or an ARIMA model, and one sub-difference sequence corresponds to one network model.

[0041] In another embodiment, the segmentation evaluation function further includes: calculating the mean of the difference values ​​for each sub-difference sequence, and calculating the variance of the means of all difference values.

[0042] The segmentation evaluation function satisfies the following relation:

[0043] , This represents the segmentation evaluation function. Sub-difference sequence The Middle Each difference value, Sub-difference sequence The mean of the differences, Sub-difference sequence The sequence length, This represents the total number of sub-difference sequences. Represents an exponential function. This represents the variance of the mean of all differences.

[0044] This embodiment considers not only within-group differences but also between-group differences by calculating the variance of the mean difference values ​​between different groups. Specifically, within-group differences reflect the consistency of temperature changes within each sub-difference sequence, while between-group differences reveal the magnitude and degree of variation between different sub-difference sequences. By calculating the variance, it is possible to quantify whether there are significant differences in the temperature change trends of different sub-difference sequences. The larger the variance, the more obvious the between-group differences and the better the segmentation effect. Conversely, a smaller variance indicates that the differences between different sub-difference sequences are more similar, resulting in a poorer segmentation effect. By simultaneously optimizing both within-group and between-group differences, the optimal segmentation point can be selected more accurately, ensuring that each sub-difference sequence can well reflect an independent and representative temperature change trend, thereby improving the accuracy and robustness of the temperature prediction model.

[0045] S3: Input the sequence to be predicted into any network model and output the corresponding temperature prediction value. Use the importance of the calculated network model as the weight, calculate the product of the weight and the temperature prediction value, and use the sum of the products of each network model as the final temperature prediction value to complete the temperature monitoring.

[0046] In one embodiment, the sequence to be predicted is input into any network model, which outputs the corresponding temperature prediction value. The temperature prediction value corresponds one-to-one with the number of network models.

[0047] A sliding window is constructed based on the length of the sequence to be predicted. The historical sequence is divided according to a preset sliding step size to obtain several windows, and the temperature sequence is contained within each window.

[0048] Each window is predicted using any network model. The mean square error of each window is calculated based on the prediction results and the actual temperature value. The mean square error corresponds one-to-one with the number of network models.

[0049] The instability of each window is calculated. The instability includes: taking any sampling time as the target time, the previous sampling time adjacent to the target time as the adjacent time, and the difference between the temperature at the target time and the temperature at the adjacent time as the temperature change rate at the target time, thus obtaining a temperature change rate sequence for any window. The cumulative difference between each temperature change rate and the mean temperature change rate within any window is calculated. The sum of 0.01 and the cumulative difference is normalized and used as the window's instability. Adding 0.01 is to avoid an instability value of 0.

[0050] Similarly, the instability of the sequence to be predicted is obtained. Based on the instability, all windows and the sequence to be predicted are clustered to obtain clusters. The windows in the clusters containing the sequence to be predicted are used as reference windows, and their importance is calculated based on the mean squared error of the reference windows. Preferably, K-means clustering can be used as the clustering algorithm.

[0051] Importance satisfies the following relation:

[0052] , Representation of network model The importance of Representation of network model In the reference window The mean square error, Indicates the total number of reference windows. This represents the total number of network models. This represents an exponential function.

[0053] In one embodiment, the first-order difference sequence of the historical sequence is divided according to a preset sliding step size to obtain several windows. In this embodiment, the window contains a temperature difference sequence, i.e., a temperature change rate sequence, from which instability can be directly calculated.

[0054] In one embodiment, the importance includes: using the first-order difference sequence of the sequence to be predicted as the temperature difference sequence, and calculating the distance between the temperature difference sequence and any sub-difference sequence.

[0055] Importance satisfies the following relation: , Representation of network model The importance of Representing temperature difference sequences and sub-difference sequences distance, This represents the total number of sub-difference sequences. This represents an exponential function.

[0056] By calculating the importance of each network model and using it as a weight, the weight of each network model is multiplied by its corresponding temperature prediction value to obtain the contribution of each model to the final prediction value. The sum of the products of all models is then used as the final temperature prediction value.

[0057] The final temperature prediction value is compared with the preset temperature threshold to complete temperature monitoring.

[0058] The system includes a processor and a memory, the memory storing computer program instructions, which, when executed by the processor, implement a method for monitoring the temperature of a load cable for new energy applications according to the first aspect of the present invention.

[0059] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

[0060] It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the scope of protection of this application. Therefore, the scope of protection of this patent application shall be determined by the appended claims.

Claims

1. A method for monitoring the temperature of load cables used in new energy applications, characterized in that, include: Obtain the temperature sequence for a preset time period, including historical sequences and sequences to be predicted; The sequence to be predicted is constructed based on the future temperature values ​​to be predicted from the current time point and historical data; Calculate the first-order difference sequence of the historical sequence, iteratively segment the first-order difference sequence, construct a segmentation evaluation function, stop iterating when the segmentation evaluation function is maximized, obtain several sub-difference sequences, construct a network model for any sub-difference sequence, one sub-difference sequence corresponds to one network model; The segmentation evaluation function satisfies the following relation: , This represents a segmentation evaluation function that reflects within-group differences. Sub-difference sequence The Middle Each difference value, Sub-difference sequence The mean of the differences, Sub-difference sequence The sequence length, This represents the total number of sub-difference sequences; The sequence to be predicted is input into any network model, which outputs the corresponding temperature prediction value. The importance of the calculated network model is used as the weight, and the product of the weight and the temperature prediction value is calculated. The sum of the products of each network model is used as the final temperature prediction value, thus completing the temperature monitoring. Importance is determined by: constructing a sliding window based on the length of the sequence to be predicted; dividing the historical sequence into several windows according to a preset sliding step size; using any network model to predict each window to obtain the mean squared error of each window; calculating the instability of each window; similarly obtaining the instability of the sequence to be predicted; clustering all windows and the sequence to be predicted based on the instability to obtain clusters; using the windows in the cluster where the sequence to be predicted belongs as reference windows; and calculating the importance based on the mean squared error of the reference windows.

2. The temperature monitoring method for a load cable for new energy applications according to claim 1, characterized in that, The segmentation evaluation function also includes: Calculate the mean of the differences for each sub-difference sequence, and calculate the variance of the mean of all differences; The segmentation evaluation function satisfies the following relation: , This represents a segmentation evaluation function that reflects differences within and between groups. Sub-difference sequence The Middle Each difference value, Sub-difference sequence The mean of the differences, Sub-difference sequence The sequence length, This represents the total number of sub-difference sequences. Represents an exponential function. This represents the variance of the mean of all differences.

3. The method for temperature monitoring of a load cable for new energy applications according to claim 1, characterized in that, The network model is a neural network or an ARIMA model.

4. The temperature monitoring method for a load cable for new energy applications according to claim 1, characterized in that, The importance satisfies the following relation: , Representation of network model The importance of Representation of network model In the reference window The mean square error, Indicates the total number of reference windows. This represents the total number of network models. This represents an exponential function.

5. The method for temperature monitoring of a load cable for new energy applications according to claim 1, characterized in that, The instability includes: Take any sampling time as the target time, take the previous sampling time adjacent to the target time as the adjacent time, and take the temperature difference between the target time and the temperature of the adjacent time as the temperature change rate of the target time to obtain the temperature change rate sequence of any window. Calculate the cumulative difference between each temperature change rate and the mean temperature change rate in any window, and normalize 0.01 with the sum of the cumulative differences as the window's instability.

6. The method for temperature monitoring of a load cable for new energy applications according to claim 1, characterized in that, Also includes: The first-order difference sequence of the historical sequence is divided according to a preset sliding step size to obtain several windows.

7. The temperature monitoring method for a load cable for new energy applications according to claim 1, characterized in that, The importance mentioned includes: The first-order difference sequence of the sequence to be predicted is used as the temperature difference sequence, and the distance between the temperature difference sequence and any sub-difference sequence is calculated. Importance satisfies the following relation: , Representation of network model The importance of Representing temperature difference sequences and sub-difference sequences distance, This represents the total number of sub-difference sequences. This represents an exponential function.

8. A temperature monitoring system for load cables used in new energy applications, characterized in that, include: A processor and a memory, wherein the memory stores computer program instructions that, when executed by the processor, implement a temperature monitoring method for a load cable for new energy applications according to any one of claims 1-7.

Images