Method and device for evaluating and predicting health degree of charging pile of two-wheeled electric vehicle in combination with large model, and medium

Abstract

The invention relates to the field of electric vehicle charging piles, in particular to a method and device for evaluating and predicting the health degree of a two-wheeled electric vehicle charging pile in combination with a large model and a medium. Multi-dimensional data related to the health of the two-wheeled electric vehicle charging pile are collected, preprocessing and feature extraction are carried out on the data, features closely related to the health state are extracted, and then health degree evaluation and prediction are carried out. Due to the fact that multi-dimensional information is collected, the problem that in the prior art, evaluation content is single can be solved, feature extraction is carried out on data of different dimensions, then residual connection is carried out, data conversion is adopted for feature extraction, and the problem that in the optimization process, due to the fact that a network is too deep, potential gradient is small is avoided. According to the method, the health degree of the charging pile is evaluated and predicted in combination with the two-wheeled electric vehicle charging pile health degree prediction large model, polling and maintenance can be performed on the charging pile in a targeted manner, the operation and maintenance cost of the two-wheeled electric vehicle charging pile is reduced, and the service life of equipment is prolonged.

Description

Technical Field

[0001] The present invention relates to the field of electric vehicle charging piles, and in particular to a method, a device and a medium for evaluating and predicting the health of a two-wheel electric vehicle charging pile in combination with a large model. Background Art

[0002] As a high-frequency device, two-wheeled electric vehicle charging piles are prone to various faults, such as corrosion of charging connectors, motherboard failures, power failures, etc. These faults will affect the user's charging experience and reduce the utilization rate of two-wheeled electric vehicle charging piles. If the maintenance of two-wheeled electric vehicle charging piles is not in place, once the equipment fails, it is often difficult to repair it in time. How to better solve practical problems with limited operation and maintenance team personnel and operation and maintenance costs has become the current problem.

[0003] The existing health assessment methods for two-wheeled electric vehicle charging piles generally have problems such as single assessment content and inaccurate understanding of complex data. Summary of the invention

[0004] In view of the defects of the prior art, the present invention provides a method, device and medium for evaluating and predicting the health of a two-wheeled electric vehicle charging pile in combination with a large model, so as to solve the problems of the prior art such as single evaluation content and inaccurate understanding of complex data.

[0005] In order to solve the technical problem, the technical solution adopted by the present invention is: a method for evaluating and predicting the health of a two-wheeled electric vehicle charging pile in combination with a large model, comprising the following steps: S01, data collection, collecting data related to the health status of the two-wheeled electric vehicle charging pile, and pre-processing the collected data; S02, model training, using the data related to the health status of the two-wheeled electric vehicle charging pile collected in step S01 to train the two-wheeled electric vehicle charging pile health prediction model; S03. Real-time prediction, including: 1. Data access: collect multi-dimensional data of two-wheeled electric vehicle charging piles in real time and input it into the health prediction model of two-wheeled electric vehicle charging piles; 2. Feature extraction: The charging pile health prediction model preprocesses and extracts features from the input operating data to identify potential problems.

[0006] During feature extraction, data of different dimensions such as current, voltage, power, temperature, fault condition, maintenance record, model, specification, installation location, and service life are input into different feature extraction layers respectively. Each feature extraction layer performs data conversion and layer normalization on the data it receives. The data conversion formula is: , in represents the input of the Lth feature extraction layer, represents the output of the Lth feature extraction layer, Represents the mapping function. This data conversion method can avoid the potential small gradient problem caused by the network being too deep during the optimization process.

[0007] The layer normalization formula is: , Among them, U and B represent the mean and variance, which are used to translate and scale the data to a standard distribution with a mean of 0 and a variance of 1. A and C are large model learning parameters, and X represents the features of the input feature extraction layer. Layer normalization stabilizes the input and output of each layer within a reasonable range.

[0008] 3. Different feature extraction layers are residually connected, and the data output by the residual connection layer is the extracted features; 4. Output the prediction results. Input the processed features into the trained large model to obtain the health prediction results: , in Indicates the probability that the charging pile is healthy, is the sigmoid function, which is used to convert linear combinations into probability values. to is the weight parameter of the large model for predicting the health of two-wheeled electric vehicle charging piles, to are input features.

[0009] Furthermore, the method further includes step S04, model evolution, which combines the manual processing results to perform feedback evolution of a large model for predicting the health of two-wheeled electric vehicle charging piles, specifically: S41, manually processed result data collection and preprocessing, collecting the manually processed result data of the two-wheeled electric vehicle charging pile, and storing it in the two-wheeled electric vehicle charging pile health prediction large model, and then preprocessing the manually processed result data; S42, extracting features from the artificial processing result data, extracting key features from the pre-processed data as input variables for feedback evolution training; S43, Feedback evolutionary training and evaluation of the large model for predicting the health of two-wheeled electric vehicle charging piles. The model is trained using the data after feature extraction, and the model parameters are adjusted to optimize the prediction performance. At the same time, the prediction accuracy and generalization ability of the trained model are evaluated; S44. The large-scale model for predicting the health of two-wheeled electric vehicle charging piles is fed back for evolution, iteration and optimization. After receiving new data and manual feedback, the model conducts continuous learning and self-optimization, and regularly evaluates the model's predictive performance to ensure that the model maintains a high level of accuracy and stability.

[0010] Furthermore, in step S44, the health index evaluation data of the two-wheeled electric vehicle charging pile is subjected to predictive analysis and directional fine-tuning in combination with the manual processing results, using a supervised task fine-tuning method, specifically: S411, annotate the manual processing results, and each two-wheel charging pile input feature is composed of a text sequence with an input length of N and the corresponding label Y; S412: Input the text sequence X into the two-wheeled electric vehicle charging pile health prediction model to obtain the hidden layer output corresponding to the last word of the last layer. , through the fully connected layer combined with the Softmax function, the label prediction result is obtained: , in For the fully connected layer function, optimize the objective function fine-tuning task for the entire data set: .

[0011] Furthermore, in step S02, the specific training process is: 1. Feature extraction and selection: by analyzing the multi-dimensional data of two-wheeled electric vehicle charging piles, the features closely related to health status are extracted; 2. Model construction: Use the current mainstream open source big model to build a prediction model, verify the prediction results through the validation set, and select a suitable big model as the original network; 3. Model training and optimization: train the model through multi-dimensional data and historical data, and continuously adjust the model parameters to improve the accuracy and generalization ability of prediction. 4. Model evaluation and iteration: Evaluate the trained model and iteratively optimize the model based on the evaluation results.

[0012] Furthermore, the data related to the health status of the two-wheeled electric vehicle charging pile includes the current, voltage, power, temperature, fault conditions, maintenance records, model, specifications, installation location, and service life of the two-wheeled electric vehicle charging pile.

[0013] Furthermore, the preprocessing includes cleaning, denoising, and normalization.

[0014] The present invention also discloses a device for evaluating and predicting the health of a two-wheeled electric vehicle charging pile in combination with a large model, comprising a processor and a memory storing program instructions, wherein the processor is configured to execute the method for evaluating and predicting the health of a two-wheeled electric vehicle charging pile in combination with a large model as described above when running the program instructions.

[0015] A storage medium stores program instructions, which, when running, execute the method for evaluating and predicting the health of a two-wheeled electric vehicle charging pile using the combined model as described above.

[0016] Beneficial effects of the present invention: The present invention collects multi-dimensional data related to the health of two-wheeled electric vehicle charging piles, pre-processes and extracts features of the data, extracts features closely related to the health status, and then performs health assessment and prediction. Because multi-dimensional information is collected, the problem of single evaluation content in the prior art can be solved. At the same time, feature extraction is performed on data of different dimensions respectively, and then residual connection is performed. In addition, feature extraction uses data conversion to avoid the problem of potential small gradients caused by excessive network depth during the optimization process, and solves the problem of inaccurate understanding of complex data. The present invention combines a large model for predicting the health of two-wheeled electric vehicle charging piles to evaluate and predict the health of charging piles, and can conduct targeted inspections and maintenance of charging piles, reduce the operating and maintenance costs of two-wheeled electric vehicle charging piles, and increase the service life of the equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a flow chart of the method described in Example 1. DETAILED DESCRIPTION

[0018] In order to be able to understand the features and technical contents of the embodiments of the present disclosure in more detail, the implementation of the embodiments of the present disclosure is described in detail below in conjunction with the accompanying drawings. The attached drawings are for reference only and are not used to limit the embodiments of the present disclosure. In the following technical description, for the convenience of explanation, a full understanding of the disclosed embodiments is provided through multiple details. However, one or more embodiments can still be implemented without these details. In other cases, to simplify the drawings, well-known structures and devices can be simplified for display.

[0019] The terms "first", "second", etc. in the embodiments of the present disclosure are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the terms used in this way can be interchanged where appropriate, so as to describe the embodiments of the present disclosure described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions.

[0020] Example 1 This embodiment discloses a method for evaluating and predicting the health of a two-wheeled electric vehicle charging pile by combining a large model, such as Figure 1 As shown, the following steps are included: S01, data collection, collecting data related to the health status of the two-wheeled electric vehicle charging pile, and pre-processing the collected data. In this embodiment, the data related to the health status of the two-wheeled electric vehicle charging pile includes the current, voltage, power, temperature, fault conditions, maintenance records, model, specifications, installation location, service life, etc. of the two-wheeled electric vehicle charging pile. The pre-processing includes cleaning and screening to remove abnormal values ​​and erroneous data.

[0021] During data collection, the rules and conditions for data collection by the two-wheeled electric vehicle charging pile health prediction big model can be defined based on actual needs and the characteristics of the two-wheeled electric vehicle charging piles, so that the two-wheeled electric vehicle charging pile health prediction big model can accurately identify and process the important data used for training.

[0022] The large model for predicting the health of two-wheeled electric vehicle charging piles collects the operating data of two-wheeled electric vehicle charging piles in real time, and stores it in a two-wheeled electric vehicle charging pile health prediction storage center after unified processing.

[0023] S02: Model training, using the data related to the health status of the two-wheeled electric vehicle charging pile collected in step S01) to train the large model for predicting the health status of the two-wheeled electric vehicle charging pile.

[0024] The specific training process is: 1. Feature extraction and selection. By analyzing the multi-dimensional data of two-wheeled electric vehicle charging piles, features closely related to the health status are extracted, such as the number of charging times, fault records, and ambient temperature.

[0025] 2. Model construction. Use the current mainstream open source big model to build a prediction model, verify the prediction results through the validation set, and select a suitable big model as the original network.

[0026] 3. Model training and optimization: Train the model through multi-dimensional data and relevant historical data, and continuously adjust the model parameters to improve the accuracy and generalization ability of prediction.

[0027] 4. Model evaluation and iteration: Evaluate the trained model, such as by measuring it through indicators such as accuracy and recall, and iterate and optimize the model based on the evaluation results.

[0028] S03. Real-time prediction, including: 1. Data access: collect multi-dimensional data of two-wheeled electric vehicle charging piles in real time and input it into the health prediction model of two-wheeled electric vehicle charging piles; 2. Feature extraction: The charging pile health prediction model preprocesses and extracts features from the input operating data to identify potential problems.

[0029] During feature extraction, data of different dimensions such as current, voltage, power, temperature, fault condition, maintenance record, model, specification, installation location, and service life are input into different feature extraction layers respectively. Each feature extraction layer performs data conversion and layer normalization on the data it receives. The data conversion formula is: , in represents the input of the Lth feature extraction layer, represents the output of the Lth feature extraction layer, Represents the mapping function. This data conversion method can avoid the potential small gradient problem caused by the network being too deep during the optimization process.

[0030] The layer normalization formula is: , Among them, U and B represent the mean and variance, which are used to translate and scale the data to a standard distribution with a mean of 0 and a variance of 1. A and C are large model learning parameters, and X represents the features of the input feature extraction layer. Layer normalization stabilizes the input and output of each layer within a reasonable range.

[0031] 3. Different feature extraction layers are residually connected, and the data output by the residual connection layer is the extracted features.

[0032] 4. Output the prediction results. Input the processed features into the trained large model to obtain the health prediction results: , in Indicates the probability that the charging pile is healthy, is the sigmoid function, which is used to convert linear combinations into probability values. to is the weight parameter of the large model for predicting the health of two-wheeled electric vehicle charging piles, to are input features.

[0033] S04, model evolution, combined with the results of manual processing, the large model for predicting the health of two-wheeled electric vehicle charging piles is fed back and evolved, specifically: S41, manually processed result data collection and preprocessing, collecting the manually processed result data of the two-wheeled electric vehicle charging pile, and storing it in the two-wheeled electric vehicle charging pile health prediction large model, and then preprocessing the manually processed result data; S42, extracting features from the artificial processing result data, extracting key features from the pre-processed data as input variables for feedback evolution training; S43, Feedback evolutionary training and evaluation of the large model for predicting the health of two-wheeled electric vehicle charging piles. The model is trained using the data after feature extraction, and the model parameters are adjusted to optimize the prediction performance. At the same time, the prediction accuracy and generalization ability of the trained model are evaluated; S44. The large-scale model for predicting the health of two-wheeled electric vehicle charging piles is fed back for evolution, iteration and optimization. After receiving new data and manual feedback, the model conducts continuous learning and self-optimization, and regularly evaluates the model's predictive performance to ensure that the model maintains a high level of accuracy and stability.

[0034] In step S44, since there is little evaluation data on the health index of two-wheeled electric vehicle charging piles in the early stage, it is necessary to combine the manual processing results to perform prediction analysis and directional fine-tuning on the evaluation data on the health index of two-wheeled electric vehicle charging piles, and a supervised task fine-tuning method is used, specifically: S411, annotate the manual processing results, and each two-wheel charging pile input feature is composed of a text sequence with an input length of N and the corresponding label Y; S412: Input the text sequence X into the two-wheeled electric vehicle charging pile health prediction model to obtain the hidden layer output corresponding to the last word of the last layer. , through the fully connected layer combined with the Softmax function, the label prediction result is obtained: , in For the fully connected layer function, optimize the objective function fine-tuning task for the entire data set: .

[0035] In this embodiment, the process of training the large model for predicting the health of two-wheeled electric vehicle charging piles is as follows: Example 2 This embodiment discloses a device for evaluating and predicting the health of a two-wheeled electric vehicle charging pile in combination with a large model, including a processor and a memory storing program instructions, and the processor is configured to execute the method for evaluating and predicting the health of a two-wheeled electric vehicle charging pile in combination with a large model as described in Example 1 when running the program instructions.

[0036] Example 3 This embodiment discloses a storage medium storing program instructions, which, when run, execute the method for evaluating and predicting the health of a two-wheeled electric vehicle charging pile in combination with a large model as described in Embodiment 1.

[0037] The above description and the accompanying drawings fully illustrate the embodiments of the present disclosure so that those skilled in the art can practice them. Other embodiments may include structural, logical, electrical, process and other changes. The embodiments represent only possible changes. Unless explicitly required, separate components and functions are optional, and the order of operation may vary. The parts and features of some embodiments may be included in or replace the parts and features of other embodiments. Moreover, the words used in this application are only used to describe the embodiments and are not used to limit the scope of protection. As used in the description in the text, unless the context clearly indicates, the singular forms of "a", "an" and "the" are intended to include plural forms as well. Similarly, the term "and / or" as used in this application refers to any and all possible combinations of one or more associated listings. In addition, when used in this application, the term "comprise" and its variants "comprises" and / or comprising refer to the presence of stated features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or groups thereof. In the absence of further restrictions, the elements defined by the sentence "including one..." do not exclude the presence of other identical elements in the process, method or device including the elements. In this article, each embodiment may focus on the differences from other embodiments, and the same and similar parts between the embodiments may refer to each other. For the methods, products, etc. disclosed in the embodiments, if they correspond to the method part disclosed in the embodiments, then the relevant parts can refer to the description of the method part.

[0038] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software may depend on the specific application and design constraints of the technical solution. The technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the embodiments of the present disclosure. The technicians may clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here.

Claims

1. A method for evaluating and predicting the health of two-wheeled electric vehicle charging piles in combination with a large model, characterized in that: The following steps are involved: S01, data collection, collecting data related to the health status of the two-wheeled electric vehicle charging pile, and pre-processing the collected data; S02, model training, using the data related to the health status of the two-wheeled electric vehicle charging pile collected in step S01 to train the two-wheeled electric vehicle charging pile health prediction model; S03. Real-time prediction, including: (1) Data access: real-time collection of multi-dimensional data of two-wheeled electric vehicle charging piles and inputting it into the two-wheeled electric vehicle charging pile health prediction model; (2) Feature extraction: The charging pile health prediction model preprocesses and extracts features from the input operation data to identify potential problems; During feature extraction, data of different dimensions are input into different feature extraction layers respectively. Each feature extraction layer performs data conversion and layer normalization on the data it receives. The data conversion formula is: ， in represents the input of the Lth feature extraction layer, represents the output of the Lth feature extraction layer, Represents the mapping function. This data conversion method can avoid the potential problem of small gradients caused by too deep a network during the optimization process. The layer normalization formula is: , Among them, U and B represent the mean and variance, which are used to translate and scale the data to a standard distribution with a mean of 0 and a variance of 1. A and C are the large model learning parameters. X represents the features of the input feature extraction layer. Layer normalization makes the input and output of each layer stable within a reasonable range. (3) Different feature extraction layers are connected by residual connection. The output data of the residual connection layer is the extracted features. (4) Output the prediction results. Input the processed features into the trained large model to obtain the health prediction results: ， in Indicates the probability that the charging pile is healthy, is the sigmoid function, which is used to convert linear combinations into probability values. to is the weight parameter of the large model for predicting the health of two-wheeled electric vehicle charging piles, to are input features.

2. The method for evaluating and predicting the health of two-wheeled electric vehicle charging piles in combination with a large model according to claim 1 is characterized in that: The step S04 is also included, model evolution, combining the manual processing results to perform feedback evolution of the large model for predicting the health of two-wheeled electric vehicle charging piles, specifically: S41, manually processed result data collection and preprocessing, collecting the manually processed result data of the two-wheeled electric vehicle charging pile, and storing it in the two-wheeled electric vehicle charging pile health prediction large model, and then preprocessing the manually processed result data; S42, extracting features from the artificial processing result data, extracting key features from the pre-processed data as input variables for feedback evolution training; S43, Feedback evolutionary training and evaluation of the large model for predicting the health of two-wheeled electric vehicle charging piles, using the data after feature extraction to train the model, adjusting the model parameters to optimize the prediction performance, and at the same time, evaluating the prediction accuracy and generalization ability of the trained model; S44. The large-scale model for predicting the health of two-wheeled electric vehicle charging piles is fed back for evolution, iteration and optimization. After receiving new data and manual feedback, the model conducts continuous learning and self-optimization, and regularly evaluates the model's predictive performance to ensure that the model maintains a high level of accuracy and stability.

3. The method for evaluating and predicting the health of two-wheeled electric vehicle charging piles in combination with a large model according to claim 2 is characterized in that: In step S44, the health index evaluation data of the two-wheeled electric vehicle charging pile is subjected to predictive analysis and directional fine-tuning in combination with the manual processing results, and a supervised task fine-tuning method is used, specifically: S411, annotate the manual processing results, and each two-wheel charging pile input feature is composed of a text sequence with an input length of N and the corresponding label Y; S412: Input the text sequence X into the two-wheeled electric vehicle charging pile health prediction model to obtain the hidden layer output corresponding to the last word of the last layer. , through the fully connected layer combined with the Softmax function, the label prediction result is obtained: , in For the fully connected layer function, optimize the objective function fine-tuning task for the entire data set: 。 4. The method for evaluating and predicting the health of two-wheeled electric vehicle charging piles in combination with a large model according to claim 1 is characterized in that: In step S02, the specific training process is: (1) Feature extraction and selection: by analyzing the multi-dimensional data of two-wheeled electric vehicle charging piles, features closely related to health status are extracted; (2) Model construction: Use the current mainstream open source big model to build a prediction model, verify the prediction results through the validation set, and select a suitable big model as the original network; (3) Model training and optimization: Train the model through multi-dimensional data and historical data, and continuously adjust the model parameters to improve the accuracy and generalization ability of prediction. (4) Model evaluation and iteration: Evaluate the trained model and iteratively optimize the model based on the evaluation results.

5. The method for evaluating and predicting the health of two-wheeled electric vehicle charging piles in combination with a large model according to claim 1 is characterized in that: The data related to the health status of the two-wheeled electric vehicle charging pile includes the current, voltage, power, temperature, fault conditions, maintenance records, model, specifications, installation location, and service life of the two-wheeled electric vehicle charging pile.

6. The method for evaluating and predicting the health of two-wheeled electric vehicle charging piles in combination with a large model according to claim 1 or 2, characterized in that: The preprocessing includes cleaning, denoising and normalization.

7. A device for evaluating and predicting the health of a two-wheeled electric vehicle charging pile in combination with a large model, comprising a processor and a memory storing program instructions, characterized in that: The processor is configured to execute the method for evaluating and predicting the health of a two-wheeled electric vehicle charging pile using the combined model as described in any one of claims 1 to 6 when running the program instructions.

8. A storage medium storing program instructions, characterized in that: When the program instructions are run, they execute the method for evaluating and predicting the health of a two-wheeled electric vehicle charging pile using the combined model as described in any one of claims 1 to 6.

Images