Large-scale battery health degree prediction method for energy storage power station

Abstract

The invention relates to the field of energy storage management, in particular to a large-scale battery health degree prediction method for an energy storage power station. Preprocessing the data in the field to obtain processed data; initializing a vector database, and importing the processed data into the vector database; designing a first Prompt, packaging the input test data into the first Prompt, and then calling the large model to predict the input test data to obtain a large model SOH predicted value; based on input test data, N SOH values are retrieved from a vector database; on the basis of the SOH predicted value of the large model and the retrieved N SOH values, optimizing the first Prompt so as to call the large model to carry out iterative prediction on the input test data until the SOH predicted value, obtained through iterative prediction, of the large model meets a preset condition; and calling the large model to output a battery health evaluation and optimization suggestion based on the integrated SOH predicted value and the second Prompt. According to the method, reliable and effective large model prediction can be carried out on the specific field of the energy storage power station with low investment cost.

Description

Technical Field

[0001] The present invention relates to the field of energy storage management, and in particular to a method for large-scale prediction of battery health in energy storage power stations. Background Art

[0002] The state of health (SOH) of a battery is one of the core indicators of battery health. The degree of battery aging is usually characterized by battery capacity or battery internal resistance, which is closely related to the working safety and stability of energy storage power stations. At present, there is no unified standard definition of battery health. The mainstream definition of battery health is: the ratio of the actual available capacity of the battery to the rated capacity. The aging of batteries, especially lithium batteries, is a long and complex process. It is affected by factors such as electrochemical reactions, charge and discharge cycles, temperature changes, and usage patterns. The relationship between these factors is complex and nonlinear, and it is difficult to accurately capture them through traditional modeling methods. Existing battery health predictions mainly include direct measurement methods, model-based prediction methods, and data-driven methods.

[0003] With the development of big data and artificial intelligence, model-based prediction methods have gradually gained attention. However, the prediction of battery health in energy storage power stations based on big model technology still faces the following challenges: First, common big models are general big models, which require knowledge in specific fields to play a big role in different fields; second, the fine-tuning of big models to knowledge in specific fields requires a large amount of computing power investment and data annotation costs in the field; third, big models are language models, which still have great difficulties in implementing accurate prediction applications; fourth, existing big models have few applications in questions and answers related to the health of batteries in energy storage power stations. Summary of the invention

[0004] In order to solve the problem that the computing power required for large-scale battery health prediction of energy storage power stations using large models and the investment cost of data annotation in the field are large, it is impossible to carry out targeted and accurate prediction and optimization evaluation of the battery health of energy storage power stations, and to carry out reliable and effective large-model prediction for specific fields of energy storage power stations, the present invention provides a method for large-scale prediction of battery health of energy storage power stations, and the method comprises the following steps: Step S1, privatize and deploy a large model; pre-process the data in the field to obtain processed data; initialize a vector database, and import the processed data into the vector database; Step S2, designing a first prompt, the first prompt is used to encapsulate the battery test data, and guide the input text of the large model to perform initial prediction to encapsulate the input test data, after the first prompt, call the large model to predict the input test data, and obtain the large model SOH prediction value; based on the input test data, retrieve N SOH values ​​from the vector database; Step S3, based on the large model SOH prediction value and the retrieved N SOH values, optimizing the first Prompt, and based on the optimized first Prompt, calling the large model to iteratively predict the input test data until the large model SOH prediction value obtained by iterative prediction meets the preset conditions; Step S4, based on the SOH prediction value of the large model that meets the preset conditions and the retrieved N SOH values, an integrated SOH prediction value is obtained; a second prompt is designed, and the second prompt is based on the integrated SOH value to guide the large model to generate input text for battery health status evaluation and optimization suggestions, and based on the integrated SOH prediction value and the second prompt, the large model is called to output battery health evaluation and optimization suggestions.

[0005] Preferably, the privatized deployment large model is specifically: Privately deploy the Qwen2.5-7B model in any system environment including ubuntu22.04, Python 3.10, PyTorch 2.1.0, and Cuda 12.1.

[0006] Preferably, the preprocessing of the data in the field to obtain processed data is specifically: Performing outlier and null value cleaning processing on the laboratory simulated working condition battery data; wherein the laboratory simulated working condition battery data includes parameter data obtained by measuring a plurality of batteries under the laboratory simulated working condition; Based on the battery capacity data contained in the laboratory simulation working condition battery data that has completed the cleaning process, several SOH values ​​are calculated; based on the calculated SOH values, the laboratory simulation working condition battery data that has completed the cleaning process is subjected to SOH label addition processing to generate feature column data; The feature column data is normalized to obtain processed data.

[0007] Preferably, the initialization vector database imports the processed data into the vector database, specifically: The Faiss vector database is initialized, and the processed data is imported into the Faiss vector database.

[0008] Preferably, the first prompt is designed, and after the input test data is encapsulated into the first prompt, the large model is called to predict the input test data to obtain the large model SOH prediction value, specifically: Design a prediction prompt to obtain test data of a target battery; based on the prediction prompt, package the test data, call the large model to predict the packaged test data, and obtain the large model SOH prediction value corresponding to the target battery.

[0009] Preferably, the retrieving N SOH values ​​from the vector database based on the input test data is specifically: Vectorize the input test data to obtain vectorized input test data; perform similarity search on the vectorized input test data and all vector data in the vector database, and obtain N SOH values ​​from the vector database; wherein the N SOH values ​​refer to the SOH values ​​contained in the first N vector data in the vector database that have the highest similarity with the vectorized input test data.

[0010] Preferably, optimizing the first Prompt based on the large model SOH prediction value and the retrieved N SOH values ​​includes: Get the SOH mean corresponding to the retrieved N SOH values; Based on the numerical difference between the large model SOH prediction value and the SOH mean value, the numerical difference is spliced ​​to the first Prompt, so as to optimize the first Prompt.

[0011] Preferably, based on the optimized first Prompt, calling the large model to iteratively predict the input test data until the large model SOH prediction value obtained by iterative prediction meets a preset condition, including: Based on the optimized first prompt, call the large model to iteratively predict the input test data to obtain a new large model SOH prediction value; determine whether the size difference between the new large model SOH prediction value and the SOH mean meets the preset conditions; If it is satisfied, the current new large model SOH prediction value is used as the large model SOH prediction value that meets the preset conditions; if it is not satisfied, the first Prompt is optimized again, and the large model is called to iteratively predict the input test data to obtain a new large model SOH prediction value, until the new large model SOH prediction value meets the preset conditions.

[0012] Preferably, based on the large model SOH prediction value that meets the preset conditions and the retrieved N SOH values, an integrated SOH prediction value is obtained, including: Get the SOH mean corresponding to the retrieved N SOH values; Then, based on the corresponding weight ratio, the SOH prediction value of the large model that meets the preset conditions and the SOH mean are weighted averaged to obtain an integrated SOH prediction value.

[0013] Preferably, a second prompt is designed, and based on the integrated SOH prediction value and the second prompt, a large model is called to output battery health evaluation and optimization suggestions, including: Design a battery health evaluation and optimization prompt, and load the battery health evaluation and optimization prompt into the big model; call the big model to process the integrated SOH prediction value and the output test data to obtain a battery health evaluation and optimization suggestion that matches the battery health evaluation and optimization prompt.

[0014] Compared with the prior art, the present invention has the following beneficial effects: First, the large model can be accurately applied to the prediction scenarios of battery core indicators such as battery health of energy storage power stations; Second, combine the big model to predict and evaluate the battery health and output optimization suggestions to achieve a closed loop in the battery health status monitoring process; Third, a retrieval enhancement method based on a large model is used to integrate battery data of different working conditions, and the battery data can be dynamically adjusted to make the prediction results more stable; Fourth, a feedback link is added to the large model prediction process to conduct differential analysis between the large model prediction results and the vector retrieval results, and to put forward improvement suggestions to ensure the reliability of battery health prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them: Figure 1 It is a flow chart of a method for large-scale prediction of battery health in an energy storage power station provided by the present invention. DETAILED DESCRIPTION

[0016] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings. It is understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for ease of description, only the parts related to the present invention rather than all structures are shown in the accompanying drawings. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0017] The terms "include" and "have" and any variations thereof in the present invention are intended to cover non-exclusive inclusions. For example, a process, method, product or device comprising a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products or devices.

[0018] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0019] See also Figure 1 As shown, the present invention provides a method for large-scale prediction of battery health of an energy storage power station, the method comprising the following steps: Step S1, privatize and deploy the large model; pre-process the data in the field to obtain processed data; initialize the vector database and import the processed data into the vector database.

[0020] Furthermore, the privatized deployment model is as follows: Privately deploy the Qwen2.5-7B model in any system environment including ubuntu22.04, Python 3.10, PyTorch 2.1.0, and Cuda 12.1.

[0021] In actual work, the base class of the large model can be selected according to actual needs. The Qwen2.5-7B model can be preferably used as the large model. This is because the Qwen2.5-7B model is a relatively new domestic large model. Its performance is superior to that of large models with the same level of parameters, and it has good support for Chinese. In the process of private deployment of large models, environmental preparation is also required to confirm that the server hardware configuration meets the requirements for large model deployment, and to install the necessary software environment, where the software environment can be but is not limited to any system software environment of ubuntu22.04, Python 3.10, PyTorch 2.1.0, and Cuda 12.1. In addition, in the process of private deployment of large models, the server hardware information can be but is not limited to: CPU: 18 vCPU AMD EPYC 9754 128-Core; GPU: RTX 4090D (24GB) 1; Memory: 60GB; and in the process of privately deploying large models, pull model-related files from HuggingFace or ModelScope.

[0022] The advantages of private deployment of big models are as follows: it can ensure that data does not leave the internal environment of the enterprise, ensuring data privacy and security; it allows enterprises to optimize the performance of models in specific scenarios according to their own needs. Through big model technology, specific domain knowledge, industry terms and data can be embedded, making big models more professional and useful and not restricted by third parties; services can be provided in offline mode to reduce external dependence and risks.

[0023] Furthermore, the data in the field is preprocessed to obtain processed data, specifically: Performing outlier and null value cleaning on the laboratory simulated working condition battery data; wherein the laboratory simulated working condition battery data includes parameter data obtained by measuring a number of batteries under laboratory simulated working conditions; Based on the battery capacity data contained in the laboratory simulation working condition battery data that has completed the cleaning process, several SOH values ​​are calculated; based on the calculated SOH values, the laboratory simulation working condition battery data that has completed the cleaning process is subjected to SOH label addition processing to generate feature column data; The feature column data is normalized to obtain the processed data.

[0024] For data in the field (i.e., data related to the operation of batteries in energy storage power stations, etc.), preprocessing of data in the field mainly includes two processes: data cleaning and data normalization. Specifically, Data cleaning includes cleaning abnormal values ​​and null values ​​of laboratory simulated working condition battery data. The laboratory simulated working condition battery data may include but is not limited to parameter data such as current, voltage, charge and discharge time, number of cycles, battery capacity, etc. obtained by measuring a certain number of batteries under laboratory simulated working conditions. By cleaning abnormal values ​​and null values ​​of the above laboratory simulated working condition battery data, the reliability of battery data can be improved, and the impact of abnormal value data and null value data on the credibility of subsequent battery health prediction can be avoided. Among them, the acquisition process of laboratory simulated working condition battery data is as follows: DEBUG: Prompt sent to model: "#prompt input " This is the discharge cycle data of lithium iron phosphate battery: Cycle number: 8 Average voltage: 3.1818v Maximum voltage: 3.4747v Minimum voltage: 1.9994v Voltage difference: 1.4753v Discharge time: 7531.0 seconds Then, based on the battery capacity data contained in the laboratory simulation working condition battery data that has completed the cleaning process, the SOH value corresponding to each battery is calculated, wherein the above SOH value is equal to the ratio between the actual battery capacity corresponding to the battery capacity data and the rated capacity of the battery, so that the battery data corresponding to each battery under the laboratory simulation working condition uniquely corresponds to an SOH value. In view of the one-to-one correspondence between each SOH value and each group of battery data under the laboratory simulation working condition, the laboratory simulation working condition battery data that has completed the cleaning process is subjected to SOH label adding processing, that is, the battery data corresponding to each battery is added with its corresponding SOH value as a label to generate feature column data. The feature column data is then normalized to obtain the processed data, wherein the above normalization processing belongs to the conventional technical means in this field and will not be described here. By normalizing the feature column data, the data range can be unified, allowing the large model to converge faster when searching for the optimal solution; it can avoid certain features dominating the calculation and ignoring the importance of other features to a certain extent; it can reduce the sensitivity of the large model to outliers; normalization compresses the feature values ​​to a fixed range (such as [0, 1]), reducing the impact of outliers on the overall data distribution.

[0025] Further, the vector database is initialized and the processed data is imported into the vector database, specifically: Initialize the Faiss vector database and import the processed data into the Faiss vector database.

[0026] Considering that Faiss is an efficient vector retrieval library designed to process massive high-dimensional vector data, the present invention selects the Faiss vector database as the vector database. The core process of vectorizing structured data is to convert the numerical, classification and time features in the data into dense vectors respectively, and then form a unified vector representation through splicing; when the vectorization is completed, Faiss can efficiently and quickly index and retrieve semantically similar structured data. The Faiss vector database is initialized and the processed data is imported into the Faiss vector database, so that the vectorized representation conversion of the processed data can be realized, so that efficient similarity retrieval can be performed in high-dimensional space. In addition, the vectorization of the processed data can be updated in real time, and new data can be dynamically added to the knowledge base. The updated data is directly vectorized and stored without retraining the model, reducing the cost of fine-tuning training.

[0027] Step S2, design the first prompt, the first prompt is used to encapsulate the battery test data and is an input text to guide the large model to perform initial prediction. After the input test data is encapsulated into the first prompt, the large model is called to predict the input test data to obtain the large model SOH prediction value; based on the input test data, N SOH values ​​are retrieved from the vector database.

[0028] Furthermore, the first prompt is designed, and after the input test data is encapsulated into the first prompt, the large model is called to predict the input test data to obtain the large model SOH prediction value, which is specifically: Design a prediction prompt to obtain the test data of the target battery; based on the prediction prompt, after packaging the test data, call the large model to predict the packaged test data to obtain the large-scale SOH prediction value corresponding to the target battery. The process of calling the large model to predict the packaged test data and obtaining the large-scale SOH prediction value is as follows: DEBUG: Full model response: "#Model response " { "model": "gwen2.5:latest" "#Model version " "created at": "2024-12-10T03:32:41.291164237Z" "#Created time " "message": { "role": "assistant" "content": "In order to predict the SOH (state of health) of a battery, we need to first understand the battery's usage and aging patterns. " "Based on the data you provided and the commonly used lithium iron phosphate battery aging model, we can use a nonlinear regression algorithm to fit the S0H value of the battery at different cycle times. ” "The given information includes: ” "-Cycle number: 8 " "-Average voltage: 3.1818V " -Maximum voltage: 3.4747V " "-Minimum voltage: 1.9994V " "-Voltage difference: 1.4753V " "-Discharge time: 7531 seconds " "Based on the common lithium iron carbonate battery aging model, we can use an exponential decay model to predict the battery's S0H. The general form of the model is: " "\\[ SOH(t)= SOH_ 0\\cdot e^{-kt}\\] " "in: " "-\\(SOH(t)\\) is the SOH value at time \\(t\\). " "-\\(SOH_0\\) is the initial S0H value (100.0%). " "-\\(k\\) is the attenuation constant, which is related to the battery aging rate. " "We can estimate the value of \\(k\\) through historical data. " "Assuming the average aging behavior of other lithium iron carbonate batteries under the same discharge conditions, we can use the known SOH value after 5000 cycles (usually about 80%) to reverse the value of \\(K\\). " "The specific calculation process is relatively complicated, but it can be simplified to fitting through known data points. " "Assume that we have obtained a suitable \\(K\\) value and applied it to the current data for prediction. After model fitting and parameter adjustment, we can get the SOH value at cycle number 8. " "Based on this information and commonly used battery aging models, we can deduce: " "```python " "import numpy as np " "#Known data points " "initial soh=100.0 #initial health " "target_soh=80.0 #Expected target health (SOH value after 5000 cycles) " "cycles_to_target_soh = 5000 " "# Calculate the decay constant k " "k = -np.log(target_soh / initial_soh) / cycles_to_target_soh " "#Current cycle number " "current_cycles = 8 " "#Predict the current SOH value " "predicted_soh =initial_soh np.exp(- current_cycles) " “``` " "After the above calculations, we can get the predicted SOH value. Assume that the calculation result is: " "\\[ \\text{Battery health}=97.0\\% \\] " "So, the final output is: " "`Battery health is: 97.0'" }, "done_reason": "stop' "#Model status " "done": true, "total_duration": 5217956007, "#Total time taken by the model to generate messages " "load_duration": 16782280, "#Total time spent loading the model " "prompt_eval_count": 285, " #Number of prompt word evaluations " "prompt_eval_duration": 19000000, "#Total time spent on prompt word evaluation " "eval_count": 578, "#The number of evaluations of the model-generated message " "eval_duration": 5168000800 "#Total time consumed for model generation message evaluation " } Large models such as the Qwen2.5-7B model are language models. When interacting with a large model, a text needs to be input, and the prompt corresponds to the text that needs to be input. The prompt is used to guide the large model to generate or complete the output of the task. The prompt is the interface between the user and the large model. By designing a reasonable prompt, the behavior of the large model can be controlled to solve different tasks.

[0029] Specifically, the functions of Prompt include: task description, that is, clearly telling the big model the task to be completed; context guidance, that is, providing relevant context information for the big model to make the output more accurate; format requirements, that is, clearly specifying the output format; behavior control, that is, guiding the direction of the big model's answer through tone and content examples. The design principles of Prompt include: clarifying the task objectives, that is, accurately conveying the task requirements to avoid ambiguity; providing context information, that is, if the task requires specific background knowledge, then include the necessary context in the Prompt; reducing openness, open prompts can easily cause the output to deviate from expectations, and restrictions should be appropriately increased; language is concise, that is, avoiding complex sentences or obscure expressions to ensure that the big model understands the tasks that need to be completed. To this end, based on the above design principles, a prediction prompt is designed, which can ensure that the big model accurately completes the corresponding prediction tasks.

[0030] In addition, the test data of the target battery (i.e., the battery whose health is to be determined) (such as the battery's current, voltage, capacity, etc.) is also obtained. That is, the prediction obtained by the above design uses Prompt as a benchmark, encapsulates the test data, and then calls the large model to predict the encapsulated test data to obtain the large model SOH prediction value corresponding to the target battery, and predicts the SOH value of the target battery from the large model level.

[0031] Furthermore, based on the input test data, N SOH values ​​are retrieved from the vector database, specifically: The input test data is vectorized to obtain the vectorized input test data; the vectorized input test data is searched for similarity with all the vector data in the vector database, and N SOH values ​​are retrieved from the vector database; wherein the N SOH values ​​refer to the SOH values ​​contained in the first N vector data in the vector database that have the highest similarity with the vectorized input test data. The process of retrieving N SOH values ​​(N=3) from the vector database is as follows:

[0032] The input test data is vectorized to obtain input test data in vector form (i.e., vectorized input test data). In this way, the input test data in vector form has the same data structure as all vector data in the vector database. At this time, the vector database can be directly searched based on the input test data in vector form. Specifically, the vectorized input test data is searched for similarity with all vector data in the vector database to obtain the similarity between the vectorized input test data and each vector data in the vector database, and then the SOH values ​​contained in the first N vector data with the highest similarity are used as the N SOH values ​​finally retrieved; wherein N can be a positive integer greater than or equal to 3.

[0033] Step S3, based on the large model SOH prediction value and the retrieved N SOH values, optimize the first Prompt, and based on the optimized first Prompt, call the large model to iteratively predict the input test data until the large model SOH prediction value obtained by iterative prediction meets the preset conditions.

[0034] Further, based on the large model SOH prediction value and the retrieved N SOH values, the first Prompt is optimized, including: Get the SOH mean corresponding to the retrieved N SOH values; Based on the numerical difference between the large model SOH prediction value and the SOH mean, the numerical difference is spliced ​​to the first prompt to optimize the first prompt.

[0035] In actual work, the arithmetic mean of N SOH values ​​is first calculated to obtain the SOH mean; then the difference between the large model SOH prediction value and the SOH mean is calculated to obtain the numerical difference between the two, and then the above numerical difference is concatenated to the first Prompt, thereby optimizing the prediction Prompt (i.e. the first Prompt) and providing effective feedback adjustment for the subsequent large model cycle prediction of battery health.

[0036] Further, based on the optimized first prompt, the large model is called to iteratively predict the input test data until the large model SOH prediction value obtained by iterative prediction meets the preset conditions, including: Based on the optimized first prompt, call the large model to iteratively predict the input test data to obtain a new large model SOH prediction value; determine whether the size difference between the new large model SOH prediction value and the SOH mean meets the preset conditions; If it is satisfied, the current new large model SOH prediction value is used as the large model SOH prediction value that meets the preset conditions; if it is not satisfied, the first Prompt is optimized again, and the large model is called to iteratively predict the input test data to obtain a new large model SOH prediction value, until the new large model SOH prediction value meets the preset conditions.

[0037] Using the optimized first Prompt, call the large model to iteratively predict the input test data to obtain a new large model SOH prediction value, and judge whether the size difference ratio between the new large model SOH prediction value and the SOH mean is less than the preset ratio threshold, that is, judge whether (new large model SOH prediction value - SOH mean) / new large model SOH prediction value is less than the preset ratio threshold (such as 2% or 5%); if it is less than the preset ratio threshold, it is determined that the size difference between the new large model SOH prediction value and the SOH mean meets the preset conditions. At this time, the current new large model SOH prediction value is used as the large model SOH prediction value that meets the preset conditions; if it is not less than the preset ratio threshold, it is determined that the size difference between the new large model SOH prediction value and the SOH mean does not meet the preset conditions. At this time, optimize the first Prompt again, and call the large model to iteratively predict the input test data to obtain a new large model SOH prediction value, until the new large model SOH prediction value meets the preset conditions.

[0038] The optimized first prompt (i.e. the optimized prediction prompt) is reloaded into the big model to add and update the data of the original big model SOH prediction value, so that it can be used in the next prediction of the big model on the input test data, to make targeted and accurate predictions and optimization evaluations of the battery health of the energy storage power station, and to make reliable and effective big model predictions for specific areas of the energy storage power station.

[0039] Step S4, based on the SOH prediction value of the large model that meets the preset conditions and the retrieved N SOH values, an integrated SOH prediction value is obtained; a second prompt is designed, and the second prompt is based on the integrated SOH value to guide the large model to generate input text for battery health status evaluation and optimization suggestions, and based on the integrated SOH prediction value and the second prompt, the large model is called to output battery health evaluation and optimization suggestions.

[0040] Furthermore, based on the large model SOH prediction value that meets the preset conditions and the retrieved N SOH values, an integrated SOH prediction value is obtained, including: Get the SOH mean corresponding to the retrieved N SOH values; Then, based on the corresponding weight ratio, the SOH prediction value and SOH mean of the large model that meet the preset conditions are weighted averaged to obtain the integrated SOH prediction value.

[0041] In actual work, the arithmetic mean of N SOH values ​​is first calculated to obtain the SOH mean. ; Then based on the corresponding weight ratio, the SOH prediction value of the large model is and SOH mean Perform weighted average processing to obtain the integrated SOH prediction value ;for example: .

[0042] The large model SOH prediction value and the SOH mean are weighted averaged, and the two SOH values ​​are combined to make the integrated SOH prediction value more stable.

[0043] Furthermore, the second prompt is designed. Based on the integrated SOH prediction value and the second prompt, the large model is called to output battery health evaluation and optimization suggestions, including: Design a battery health evaluation and optimization prompt, and load the battery health evaluation and optimization prompt into the big model; call the big model to process the integrated SOH prediction value and output test data to obtain battery health evaluation and optimization suggestions that match the battery health evaluation and optimization prompt.

[0044] When the integrated SOH prediction value of the test battery is greater than 95: DEBUG: Full model response: "#Model response " { "model": "qwen2.5:latest", "#Model version " "created at": "2024-12-10T07:25:34.4285482332", "#Creation time " "message": { "role": "assistant", "content": "The current SOH health of the lithium iron phosphate battery is 96.56, indicating that the overall health of the battery is good and close to full load operation. " "For the use scenarios of energy storage power plants, the following optimization measures are recommended: " "1. Regular maintenance : Perform battery performance test and cleaning once a quarter. " "2. Temperature control : Keep the battery operating in a suitable temperature range (about 20-35℃) and avoid overheating or overcooling. " "3. Equalizing Charge : Adopt intelligent management system to balance the voltage of each single battery to prevent overcharging or discharging. " "4. Load Management : Optimize the load distribution of the energy storage system and reduce large current shocks. " "Summary: The current health of lithium iron carbonate battery S0H is good. For energy storage power plant application scenarios, attention should be paid to daily maintenance and fine management of temperature and load to ensure long-term stable operation." }, "done_reason": "stop", "#Model status " "done": true, "total_duration": 1466557467, "#Total time taken by the model to generate messages " "load_duration": 16705462, "#Total time spent loading the model " "prompt_eval_count": 196, " #Number of prompt word evaluations " "prompt_eval_duration": 19000000, "#Total time spent on prompt word evaluation " "eval_count": 166, "#The number of evaluations of the model-generated message " "eval_duration": 1423000000 "#Total time consumed for model to generate message " } When the integrated SOH prediction value of the test battery is greater than or equal to 80 and less than or equal to 95: DEBUG: Full model response: "#Model response " { "model": "qwen2.5:latest", "#Model version " "created at": "2024-12-10T07:47:23.817322235Z", " # creation time " "message": { "role": "assistant", "content": "The current SOH health of lithium iron phosphate batteries is 90.08, indicating that the overall health of the battery is good, but there is still a certain risk of degradation. " "For the use scenarios of energy storage power plants, the following optimization measures are recommended: " "1. Regular maintenance : Perform a deep discharge test every quarter to ensure stable battery performance. " "2. Temperature control : Keep the battery working in an environment of about 25℃ to avoid high temperature affecting battery life. " "3. Equalizing Charge : Use the smart balancer to regularly balance power to prevent local overheating. " "4. Load Management : Reasonably arrange the charging and discharging strategy to avoid frequent high-current charging and discharging. " "Summary: For lithium iron phosphate batteries in energy storage power plants, scientific maintenance and reasonable management can effectively extend the battery life and improve system reliability." }, "done_reason": "stop", "#Model status " "done": true, "total_duration": 1422452249, "#Total time taken by the model to generate messages " "load_duration": 36079918, "#Total time to load the model " "prompt_eval_count": 196, " #Number of prompt word evaluations " "prompt_eval_duration": 18000000, "#Total time spent on prompt word evaluation " "eval_count": 153, "#The number of evaluations of the model to generate messages " "eval_duration": 1355000000 "#Total time consumed for model to generate message " } When the integrated SOH prediction value of the test battery is less than 80: DEBUG: Full model response: "#Model response " { "model": "qwen2.5:latest", "#Model version " "created at": "2024-12-10T09:26:02.150231452Z", " # creation time " "message": { "role": "assistant", "content": "According to the data provided, after 1646 cycles, the health status (SOH) of the lithium iron phosphate battery was 79.78, which is lower than the ideal threshold of 80, indicating that the battery has obviously aged. " "In energy storage power plant applications, this level of battery performance degradation may affect the reliability and effectiveness of the system. " "For the use scenarios of energy storage power plants, the following optimization measures are recommended: " "1. Regular monitoring and evaluation : Continue to track battery performance to ensure any further changes are detected promptly. " "2. Balanced charging management : Implement intelligent charging and discharging strategies to reduce voltage differences and thermal inequality. " "3. Temperature control : Optimize the battery pack cooling system to maintain a suitable operating temperature range. " "Summary: Since the SOH is below 80, it is obviously aged. It is recommended to replace the battery as soon as possible to ensure the stability and efficiency of the energy storage system." }, "done_reason": "stop", "#Model status " "done": true, "total_duration": 1508520017, "#Total time taken by the model to generate messages " "load_duration": 35971127, "#Total time to load the model " "prompt_eval_count": 256, " #Number of prompt word evaluations " "prompt_eval_duration": 12000000, "#Total time spent on prompt word evaluation " "eval_count": 164, "#The number of evaluations of the model to generate messages " "eval_duration": 1446000000 "#Total time consumed for model to generate message " } Based on the above design principles, the battery health evaluation and optimization prompt is designed to ensure that the large model can accurately complete the corresponding battery health evaluation and optimization suggestion tasks, and the battery health evaluation and optimization prompt is loaded into the large model to ensure that the large model can quickly and efficiently perform the battery health evaluation and optimization suggestion tasks. Then call the large model to process the integrated SOH prediction value and output test data to obtain the battery health evaluation and optimization suggestions that match the battery health evaluation and optimization prompt; wherein, the battery health evaluation and optimization suggestions can be, but are not limited to, a summary of battery optimization suggestions within 200 words.

[0045] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by adding a necessary general hardware platform, and of course can also be implemented by combining hardware and software. Based on such an understanding, the above technical solution can essentially or in other words be embodied in the form of a computer product, and the present invention can be implemented in the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0046] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it, and other embodiments may also be used. Although the present invention has been described in detail with reference to the aforementioned embodiments, ordinary technicians in this field should understand that they can still modify the technical solutions recorded in the aforementioned embodiments, or replace some of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting battery health on a large scale for energy storage power stations, characterized in that: The method comprises the following steps: Step S1, privatize and deploy a large model; pre-process the data in the field to obtain processed data; initialize a vector database, and import the processed data into the vector database; Step S2, designing a first prompt, the first prompt is used to encapsulate the battery test data, and guide the input text of the large model to perform initial prediction to encapsulate the input test data, after the first prompt, call the large model to predict the input test data, and obtain the large model SOH prediction value; based on the input test data, retrieve N SOH values ​​from the vector database; Step S3, based on the large model SOH prediction value and the retrieved N SOH values, optimizing the first Prompt, and based on the optimized first Prompt, calling the large model to iteratively predict the input test data until the large model SOH prediction value obtained by iterative prediction meets the preset conditions; Step S4, based on the SOH prediction value of the large model that meets the preset conditions and the retrieved N SOH values, an integrated SOH prediction value is obtained; a second prompt is designed, and the second prompt is based on the integrated SOH value to guide the large model to generate input text for battery health status evaluation and optimization suggestions, and based on the integrated SOH prediction value and the second prompt, the large model is called to output battery health evaluation and optimization suggestions.

2. The method according to claim 1, characterized in that The privatized deployment model is specifically as follows: Privately deploy the Qwen2.5-7B model in any system environment including ubuntu22.04, Python 3.10, PyTorch 2.1.0, and Cuda 12.

1.

3. The method according to claim 1, characterized in that The preprocessing of the data in the field to obtain processed data is specifically as follows: Performing outlier and null value cleaning processing on the laboratory simulated working condition battery data; wherein the laboratory simulated working condition battery data includes parameter data obtained by measuring a plurality of batteries under the laboratory simulated working condition; Based on the battery capacity data contained in the laboratory simulation working condition battery data that has completed the cleaning process, several SOH values ​​are calculated; based on the calculated SOH values, the laboratory simulation working condition battery data that has completed the cleaning process is subjected to SOH label addition processing to generate feature column data; The feature column data is normalized to obtain processed data.

4. The method according to claim 1, characterized in that: The initialization vector database imports the processed data into the vector database, specifically: The Faiss vector database is initialized, and the processed data is imported into the Faiss vector database.

5. The method according to claim 1, characterized in that The first prompt is designed, and after the input test data is encapsulated into the first prompt, the large model is called to predict the input test data to obtain the large model SOH prediction value, which is specifically: Design a prediction prompt to obtain test data of a target battery; based on the prediction prompt, package the test data, call the large model to predict the packaged test data, and obtain the large model SOH prediction value corresponding to the target battery.

6. The method according to claim 1, characterized in that The step of retrieving N SOH values ​​from the vector database based on the input test data is specifically: Vectorize the input test data to obtain vectorized input test data; perform similarity search on the vectorized input test data and all vector data in the vector database, and obtain N SOH values ​​from the vector database; wherein the N SOH values ​​refer to the SOH values ​​contained in the first N vector data in the vector database that have the highest similarity with the vectorized input test data.

7. The method according to claim 1, characterized in that Optimizing the first Prompt based on the large model SOH prediction value and the retrieved N SOH values, including: Get the SOH mean corresponding to the retrieved N SOH values; Based on the numerical difference between the large model SOH prediction value and the SOH mean value, the numerical difference is spliced ​​to the first Prompt, so as to optimize the first Prompt.

8. The method according to claim 7, characterized in that Based on the optimized first prompt, calling the large model to iteratively predict the input test data until the large model SOH prediction value obtained by iterative prediction meets the preset conditions, including: Based on the optimized first prompt, call the large model to iteratively predict the input test data to obtain a new large model SOH prediction value; determine whether the size difference between the new large model SOH prediction value and the SOH mean meets the preset conditions; If it is satisfied, the current new large model SOH prediction value is used as the large model SOH prediction value that meets the preset conditions; if it is not satisfied, the first Prompt is optimized again, and the large model is called to iteratively predict the input test data to obtain a new large model SOH prediction value, until the new large model SOH prediction value meets the preset conditions.

9. The method according to claim 1, characterized in that: Based on the large model SOH prediction value that meets the preset conditions and the retrieved N SOH values, the integrated SOH prediction value is obtained, including: Get the SOH mean corresponding to the retrieved N SOH values; Then, based on the corresponding weight ratio, the SOH prediction value of the large model that meets the preset conditions and the SOH mean are weighted averaged to obtain an integrated SOH prediction value.

10. The method according to claim 9, characterized in that Design a second prompt, based on the integrated SOH prediction value and the second prompt, call the large model to output battery health evaluation and optimization suggestions, including: Design a battery health evaluation and optimization prompt, and load the battery health evaluation and optimization prompt into the big model; call the big model to process the integrated SOH prediction value and the output test data to obtain a battery health evaluation and optimization suggestion that matches the battery health evaluation and optimization prompt.

Images