Automobile air conditioner refrigerant slow leakage fault prediction system and method based on big data

Abstract

The invention discloses an automobile air conditioner refrigerant slow leakage fault prediction system and method based on big data, and relates to the technical field of fault prediction. Comprising an arrangement module, a feature module, an acquisition module, a prediction module and an alarm module; the arrangement module is used for screening and extracting historical data of refrigerant leakage of a target automobile air conditioner model from a database by using a big data technology, and preprocessing the historical data; according to the technical key points, historical data of refrigerant leakage of a target automobile air conditioner model is accurately collected by utilizing a big data technology, processing is carried out based on the historical data of the target automobile air conditioner model, a characteristic category capable of showing refrigerant leakage is extracted, the characteristic category is further processed, and the characteristic category of the target automobile air conditioner model is extracted. Therefore, the fault prediction of the slow leakage of the refrigerant of the automobile air conditioner can be conveniently, accurately and quickly realized subsequently, the use effect is good, and the method has a good use prospect.

Description

Technical Field

[0001] The present invention relates to the technical field of fault prediction, and specifically to an automotive air-conditioning refrigerant slow-leakage fault prediction system and method based on big data. Background Art

[0002] With the continuous improvement of the social development level, the use of refrigeration equipment can provide a good environment for people's daily production and life. Among them, the refrigerant is the main raw material for realizing air-conditioning refrigeration, with strong thermodynamic properties and can also achieve energy-saving construction. Specifically in the automotive industry, refrigerant leakage in some automotive air-conditioning equipment during abnormal operation has become the main obstacle to the popularization and application of automotive air-conditioning refrigerants.

[0003] During the long-term operation of automotive air conditioners, the connection positions of different components may become loose or cracked due to vibration and corrosion. Therefore, the interior of some air conditioners with a long operation time is always in a high-pressure state, which is the main factor leading to refrigerant leakage. At present, most refrigerants are flammable. The leaked refrigerant gradually accumulates in a small space, making it easy to cause fires and explosions, and the volatilized refrigerant will also have a certain impact on human health. Therefore, at present, implementing refrigerant leakage and fault prediction has become the focus of air-conditioning safety operation and maintenance management.

[0004] Currently, in a Chinese patent with the patent application number: "CN110503217B", a method, device, equipment and system for predicting slow leakage of air-conditioning refrigerant are disclosed. It records obtaining air-conditioning operation parameters and train environment data and performing filtering processing to obtain current data; obtaining historical data, and using the historical data and the current data to obtain parameter factors; inputting the current data, the historical data and the parameter factors into a slow-leakage prediction Bayesian model to obtain a fault probability value; judging whether the fault probability value is within the slow-leakage fault prediction interval of the air-conditioning refrigerant; if so, sending a slow-leakage fault warning of the air-conditioning refrigerant; this method solves the safety hazards brought by slow leakage of train air-conditioning refrigerant, avoids unplanned absenteeism of trains due to slow-leakage faults of air-conditioning refrigerant, and at the same time does not require maintenance personnel to be on standby at the maintenance station at any time, thus saving manpower and material resources; in addition, the present invention also provides a device, equipment, system and computer-readable storage medium for predicting slow leakage of air-conditioning refrigerant, which also has the above beneficial effects.

[0005] However, during the implementation of the above technical solution, it is found that there are at least the following technical problems:

[0006] The solution of this patent is essentially to obtain the operating parameters of the air conditioner and the train environment data, perform filtering processing to obtain the current data, obtain historical data, and use the historical data and the current data to obtain parameter factors. Then, input the current data, historical data, and parameter factors into the slow-leakage prediction Bayesian model to obtain the failure probability value, and determine whether the failure probability value is within the slow-leakage failure prediction interval of the air conditioner refrigerant.

[0007] This method is based on the relatively unified train operation environment and unified train vehicles, and its operation years are long. It can directly analyze based on the historical data of the train to determine whether there is a refrigerant leak in the air conditioner. However, for general cars, the specifications of car air conditioners are different, and the data and related reactions when leaks occur are completely different. Using a unified model cannot handle this situation at all. Moreover, the car air conditioner is close to the drive structure, and the harm of refrigerant leakage is relatively large. Therefore, it is necessary to predict the slow leakage of the car air conditioner refrigerant in a timely and accurate manner. For this reason, a slow-leakage failure prediction system and method for car air conditioners based on big data are provided. Summary of the Invention

[0008] (1) Technical Problems to be Solved

[0009] In view of the deficiencies of the prior art, the present invention provides a slow-leakage failure prediction system and method for car air conditioners based on big data. It uses big data technology to accurately collect the historical data of refrigerant leakage of the target car air conditioner model, processes the historical data of the target car air conditioner model, extracts the characteristic categories that can show refrigerant leakage, and further processes the characteristic categories to obtain a comprehensive analysis value that can significantly judge whether the refrigerant leaks, so as to facilitate the subsequent accurate and rapid prediction of the slow-leakage failure of the car air conditioner refrigerant, with good use effects and good application prospects, and solves the problems raised in the background technology.

[0010] (2) Technical Solutions

[0011] To achieve the above objectives, the present invention is realized through the following technical solutions:

[0012] A slow-leakage failure prediction system for car air conditioners based on big data includes a sorting module, a feature module, a collection module, a prediction module, and an alarm module:

[0013] Sorting module: Use big data technology to screen and extract the historical data of refrigerant leakage of the target car air conditioner model from the database and perform preprocessing;

[0014] Feature module: Analyze the preprocessed data using a feature selection algorithm, generate a category importance ranking, select the top N groups of categories, calculate the comprehensive analysis value, and summarize and construct a historical data set;

[0015] Collection module: Obtain the automotive air conditioning data in real time and calculate the real-time comprehensive analysis value according to the calculation logic of the feature module;

[0016] Prediction module: Obtain the automotive air conditioning operation data, input the air conditioning operation data and the real-time comprehensive analysis value into a pre-trained convolutional neural network model, obtain the quantitative evaluation percentage of cognitive or neurological function, and obtain the refrigerant leakage percentage data for evaluating the degree of refrigerant leakage based on the cognitive or neurological function;

[0017] Alarm module: Obtain the refrigerant leakage degree data, calculate the leakage rate based on the change of the leakage degree over time, estimate the service life and risk value in sequence, and then compare the service life and risk value with the preset rules to execute the corresponding warning strategy.

[0018] Furthermore, the historical refrigerant leakage data of the automotive air conditioning includes, but is not limited to, the automotive ambient temperature, condenser inlet pressure, condenser inlet temperature, condenser outlet pressure, condenser outlet temperature, compressor suction temperature, compressor discharge temperature, compressor suction pressure, compressor discharge pressure, compressor power, evaporator inlet pressure, evaporator inlet temperature, evaporator outlet pressure, evaporator outlet temperature, and refrigerant pipeline vibration parameters at the time of refrigerant leakage.

[0019] Furthermore, the steps for preprocessing the historical refrigerant leakage data are as follows:

[0020] Compare the identification codes of the data and remove duplicate data;

[0021] Use the mean filling method to process missing values;

[0022] Integrate the processed data, and the data of each category forms a complete data set.

[0023] Furthermore, use the feature selection algorithm to analyze the preprocessed data, and the steps for generating the importance ranking of categories are as follows:

[0024] Use the Z-score standardization method to standardize the data in the data set, and then use the min-max normalization method to normalize the data;

[0025] Use the Pearson correlation coefficient method to analyze the data and calculate the correlation coefficients between all categories and the degree of refrigerant leakage;

[0026] Obtain the category with the largest absolute value of the correlation coefficient with the degree of refrigerant leakage, mark it as the key category, and then calculate the correlation coefficients between the key category and other categories in sequence;

[0027] Compare the correlation coefficients between the key category and other categories with the preset standard interval, and delete the categories not within the standard interval;

[0028] Arrange the remaining categories according to the magnitude of the correlation coefficient with the refrigerant leakage degree to generate the category importance ranking.

[0029] Furthermore, the steps for selecting the top N groups of categories and calculating the comprehensive analysis value are as follows:

[0030] Select the correlation coefficient between the top N groups of categories and the refrigerant leakage degree and the correlation coefficient between the categories.

[0031] Calculate the evaluation value PGZ using the correlation coefficient, and then calculate the adjustment coefficient PGs based on the evaluation value. The specific formulas are as follows:

[0032]

[0033] In the formula, PGs i is the adjustment coefficient of the i-th category, PGZ i is the evaluation value of the i-th category, Qtxsi j is the correlation coefficient between the i-th category and the j-th category, QZ1 and QZ2 are weight coefficients, QZ1 < QZ2, QZ1 and QZ2 are obtained by expert evaluation and scoring, XGxsi is the correlation coefficient between the i-th category and the refrigerant leakage degree, A is a constant coefficient, 0.8 ≤ A ≤ 1;

[0034] Adjust the category data using the adjustment coefficient PGs, and then superimpose the adjusted category data to obtain the comprehensive analysis value.

[0035] Furthermore, the training method of the convolutional neural network model is as follows:

[0036] Divide the historical data set into a leakage degree training set and a leakage degree test set; the leakage degree training data includes leakage degree reflection feature data and its corresponding quantization score, and the leakage degree reflection feature data includes the refrigerant leakage percentage and the comprehensive analysis value;

[0037] Construct a convolutional neural network for performing regression tasks, use the leakage degree reflection feature data in the leakage degree training set as the input of the convolutional neural network, and use the refrigerant leakage percentage of the leakage degree reflection feature data as the output of the convolutional neural network, and train the convolutional neural network to obtain an initial convolutional neural network;

[0038] Use the leakage degree test set to verify the initial convolutional neural network model, and output the initial convolutional neural network with a preset test accuracy greater than or equal to as the trained convolutional neural network model.

[0039] Further, compare the refrigerant leakage percentage data with the leakage determination percentage threshold. If the refrigerant leakage percentage data is less than the leakage determination percentage threshold, it is determined that there is no slow leakage of the automotive air-conditioning refrigerant, and no treatment is required. If the refrigerant leakage percentage data is not less than the leakage determination percentage threshold, it is determined that there is a slow leakage of the automotive air-conditioning refrigerant. At this time, calculate the leakage rate based on the change of the leakage degree over time. The specific steps are as follows:

[0040] Collect K groups of continuously collected refrigerant leakage percentage data, and calculate the difference data of the refrigerant leakage percentage between adjacent collection periods;

[0041] Conduct statistical analysis on the K - 1 groups of difference data to obtain the average value Cp, and obtain the maximum value Cmax and the minimum value Cmin of the K - 1 groups of difference data;

[0042] Calculate the difference ratios between Cmax and Cmin and Cp respectively, and compare the two groups of difference ratios with the preset fluctuation threshold. If both groups of difference ratios are less than the fluctuation threshold, it is determined that the leakage rate is the average value Cp. If there is any group of difference ratios not less than the fluctuation threshold, it is determined that the leakage rate is the maximum value Cmax.

[0043] Further, the steps to estimate the service life are as follows:

[0044] Obtain the refrigerant leakage percentage and leakage rate in the current state;

[0045] Compare the leakage rate with the preset leakage rate threshold level, and adjust the leakage rate based on the correction coefficient corresponding to the leakage rate threshold level;

[0046] Calculate the difference between the critical refrigerant usage percentage and the refrigerant leakage percentage, and estimate the service life using this difference and the adjusted leakage rate.

[0047] Further, the steps to estimate the risk value are as follows:

[0048] Obtain the refrigerant leakage percentage, service life, and leakage rate threshold level corresponding to the leakage rate in the current state;

[0049] Use the calculation method of weighted summation to process the refrigerant leakage percentage, service life, and leakage rate threshold level corresponding to the leakage rate to obtain the risk value.

[0050] Further, a method for predicting slow leakage faults of automotive air-conditioning refrigerants based on big data includes the following steps:

[0051] Use big data technology to screen and refine the historical data of refrigerant leakage of the target automotive air-conditioning model from the database, and perform preprocessing;

[0052] Use a feature selection algorithm to analyze the preprocessed data, generate a ranking of category importance, select the top N groups of categories, calculate the comprehensive analysis value, and summarize and construct a historical dataset;

[0053] Obtain automotive air conditioning data in real time and calculate the real-time comprehensive analysis value;

[0054] Obtain the operating data of the automotive air conditioner, input the air conditioner operating data and the real-time comprehensive analysis value into a pre-trained convolutional neural network model, obtain the quantitative evaluation percentage of cognitive or neurological functions, and obtain the refrigerant leakage percentage data for evaluating the degree of refrigerant leakage based on the evaluation of cognitive or neurological functions;

[0055] Obtain the refrigerant leakage degree data, calculate the leakage rate based on the change of the leakage degree over time, estimate the service life and risk value in sequence, and then compare the service life and risk value with the preset rules to execute the corresponding warning strategy.

[0056] (III) Beneficial effects

[0057] The present invention provides a system and method for predicting slow refrigerant leakage faults in automotive air conditioners based on big data, having the following

[0058] Beneficial effects:

[0059] The present invention provides a system and method for predicting slow refrigerant leakage faults in automotive air conditioners based on big data, which uses big data technology to accurately collect historical data of refrigerant leakage of the target automotive air conditioner model, and processes the historical data of the target automotive air conditioner model to extract feature categories that can show refrigerant leakage, and further processes the feature categories to obtain a comprehensive analysis value that can significantly judge whether the refrigerant leaks, so as to facilitate the subsequent accurate and rapid prediction of slow refrigerant leakage faults in automotive air conditioners, with good use effects and good application prospects.

[0060] The present invention provides a system and method for predicting slow refrigerant leakage faults in automotive air conditioners based on big data, which further analyzes on the basis of the feature categories that show refrigerant leakage, and provides a more accurate and comprehensive tool for evaluating the degree of refrigerant leakage by analyzing and mining the complex data relationship between the comprehensive analysis value and cognitive or neurological functions. Compared with traditional methods, the present invention can accurately realize the degree of refrigerant leakage, and further realize the prediction of service life and the risk value of use, laying a foundation for users to make reasonable choices, so as to provide a more suitable use plan and help users better use automotive air conditioners, having good application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 It is a partial flow block diagram of a system for predicting slow refrigerant leakage faults in automotive air conditioners based on big data in the present invention;

[0062] Figure 2 This is the flowchart of a method for predicting the slow refrigerant leakage fault of an automotive air conditioner based on big data in the present invention. Detailed implementation manner

[0063] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0064] With the development of Internet technology in recent years, big data technology is a set of technical systems for extracting, processing, and analyzing large-scale data, aiming to mine value from massive data and provide support for decision-making. By collecting its historical operation data and analyzing the changes in the data, data comparison can help determine the current specific leakage status of the equipment, which has a certain promoting effect on improving the safety and stability of air conditioner operation.

[0065] In short, studying the slow refrigerant leakage fault of an automotive air conditioner can help relevant technical personnel understand the degree of slow refrigerant leakage and the danger of leakage, quickly formulate corresponding safety usage plans for relevant users, and effectively improve the safety during the use of refrigerant leakage.

[0066] The specific implementation plan of the solution is as follows:

[0067] 1. Instrument selection

[0068] The core of this solution lies in the innovation of the refrigerant leakage identification method, which is a data processing method proposed on the basis of solving the current problems in processing slow refrigerant leakage data; all the data of the present invention comes from the relevant data of refrigerant leakage recorded in the existing automotive database, and relevant data is directly obtained through Internet of Things data.

[0069] In addition, most of the existing automotive air conditioners have the ability to detect in this regard. Therefore, there is no need to add corresponding sensors to collect data. Therefore, this system does not need to rectify the existing automotive air conditioner data acquisition system when in use.

[0070] The system on which this patent is based is a server and corresponding data communication equipment, without the need for other hardware equipment. Therefore, the implementation cost of this system is relatively low.

[0071] 2. Data processing

[0072] The entire data processing process is as Figure 1 shown, and the specific content is as follows:

[0073] 2.1 Data collation

[0074] When in use, the implementation of all systems must be based on accurate data. Only based on accurate data can further analysis be carried out and accurate predictions can be achieved. Therefore, the basis for the implementation of this system is the need for accurate data, and the process of collecting data is based on the sorting module.

[0075] Sorting module: Use big data technology to filter and extract historical data of refrigerant leakage of target automobile air-conditioning models from the database and perform pre-processing;

[0076] Big data technology, as one of the core driving forces of today's digital age, has demonstrated remarkable capabilities based on advanced distributed computing frameworks and efficient algorithms. It can process massive amounts of data in a high-speed parallel manner, whether it is structured data, such as table data in a database, or unstructured data, such as text files, images, videos, etc.

[0077] Big data technology relies on distributed computing frameworks and efficient algorithms, has the ability to process massive amounts of data in parallel at high speed, and can clean and integrate various structured and unstructured data in a very short time.

[0078] In the data processing process, the first step is data cleaning. Since raw data often contains various noises, erroneous data, and duplicate information, data cleaning is particularly critical. Big data technology uses specific algorithms and tools to quickly identify and eliminate these bad data, laying a good foundation for subsequent analysis and application. The next step is data integration, which integrates data from different data sources and formats into an organic whole, making it easier to process in a unified manner.

[0079] For the target automobile air-conditioning model, relying on the powerful mining and analysis capabilities of big data technology, the historical data of refrigerant leakage can be accurately screened and refined from the massive database. Not only can the data context be quickly sorted out, but also the data features can be deeply mined with extremely high accuracy.

[0080] For example, leakage frequency distribution, time-varying trends, leakage patterns under different working conditions, etc. These deeply refined data provide solid and powerful support for subsequent research and decision-making on automobile air-conditioning fault prediction and maintenance strategy formulation, and effectively improve the scientificity and accuracy of automobile air-conditioning leakage identification.

[0081] The historical data of refrigerant leakage in automotive air conditioners includes, but is not limited to, the ambient temperature of the vehicle, the inlet pressure of the condenser, the inlet temperature of the condenser, the outlet pressure of the condenser, the outlet temperature of the condenser, the suction temperature of the compressor, the discharge temperature of the compressor, the suction pressure of the compressor, the discharge pressure of the compressor, the compressor power, the inlet pressure of the evaporator, the inlet temperature of the evaporator, the outlet pressure of the evaporator, the outlet temperature of the evaporator, and the vibration parameters of the refrigerant pipeline.

[0082] The steps for preprocessing the historical data of refrigerant leakage are as follows:

[0083] Compare the identification codes of the data and remove duplicate data;

[0084] Read the historical data file of refrigerant leakage and load the data into the corresponding data processing software. The data processing software identifies the identification code column in the data. The identification code is generally a field that can uniquely identify a data record, such as a timestamp. Therefore, it can be determined whether there are duplicate data. In actual use, there are cases of data packet loss. Then, to ensure data integrity, data is continuously transmitted, so there will be a situation where data is widely repeated at a single time point. To reduce the subsequent data comparison and processing volume, duplicate data needs to be deleted.

[0085] Use the mean filling method to handle missing values;

[0086] The collected data is historical data for the past three years. Due to the developed Internet of Things technology, the collected data is relatively complete. Therefore, there are fewer missing values, but missing values still need to be processed. For leakage, the most reasonable and simple solution is to use the mean filling method for processing.

[0087] The mean filling method is a commonly used method for handling missing values. It calculates the mean of non-missing values and fills this mean into the missing value position. This method is applicable to situations where the data distribution is relatively uniform and the number of missing values is small. Regarding the mean filling method, it belongs to the prior art and is often used in data processing. Therefore, no further description will be given.

[0088] When in use, the median method or the mode method can also be used for filling.

[0089] Integrate the processed data, and the data of each category forms a complete data set.

[0090] The category is the data category described in the historical data of refrigerant leakage in automotive air conditioners. Store the data of each category in a dictionary, where the key of the dictionary is the category name and the value is the corresponding table. In this way, the data of each category forms a complete data set and can be further used for analysis.

[0091] 2.2. Feature Analysis

[0092] During use, due to different vehicles, different air conditioner models are used, and different air conditioners often have different faults. Therefore, it is impossible to evaluate all air conditioners in the same way, and targeted treatment must be carried out. To ensure the accuracy of fault analysis, it is necessary to analyze and judge air conditioners individually, and this process is based on the feature module.

[0093] Feature module: Use the feature selection algorithm to analyze the preprocessed data, generate the category importance ranking, select the top N groups of categories, calculate the comprehensive analysis value, and summarize and construct the historical data set;

[0094] The steps of using the feature selection algorithm to analyze the preprocessed data to generate the category importance ranking are as follows:

[0095] Use the Z-score standardization method to standardize the data in the data set, and then use the min-max normalization method to normalize the data;

[0096] Using the Z-score standardization method to standardize the data in the data set means for each data point X in the data set i Through the formula In the formula, μ is the mean of the data in the data set, and σ is the standard deviation of the data in the data set. This step can transform the data into a standard normal distribution with a mean of 0 and a standard deviation of 1, effectively eliminating the differences in dimension and order of magnitude between different variables and improving the comparability of the data.

[0097] Using the min-max normalization method to normalize the data means for each data point X in the data set i Through the formula for conversion, where X max and X min are the maximum and minimum values of the data in the data set respectively. After this step, the data can be mapped into the interval [0, 1], further optimizing the distribution characteristics of the data and laying a foundation for subsequent analysis.

[0098] Use the Pearson correlation coefficient method to analyze the data and calculate the correlation coefficients between all categories and the degree of refrigerant leakage;

[0099] The calculation formula of the Pearson correlation coefficient method is as follows:

[0100]

[0101] In the formula, X i and Y i are the corresponding values of two variables respectively, and are the means of two variables respectively, n is the data of the variables. Through the above calculation formula, the strength and direction of the linear correlation between the two variables can be calculated, and the value range is between [-1, 1]. The correlation coefficients between all categories and the degree of refrigerant leakage are calculated.

[0102] When in use, schemes such as Spearman rank correlation coefficient, Kendall rank correlation coefficient, and partial correlation coefficient can also be adopted for calculation.

[0103] Obtain the category with the largest absolute value of the correlation coefficient related to the degree of refrigerant leakage, and mark it as the key category. The key category is the optimal category. Therefore, based on this, analysis is carried out, and then corresponding leakage prediction categories are formulated based on this, which can improve the accuracy of leakage prediction.

[0104] The fault prediction of slow refrigerant leakage in automotive air conditioners is to improve the discovery time of slow refrigerant leakage and reduce the harm caused by slow refrigerant leakage in automotive air conditioners.

[0105] Then, calculate the correlation coefficients between the key category and other categories in turn;

[0106] Compare the correlation coefficients between the key category and other categories with the preset standard interval, and delete the categories that are not within the standard interval;

[0107] For example: If there are 4 groups of other categories and the preset standard interval is (-0.4, 0.4), and the correlation coefficients between the key category and other categories are 0.3, 0.1, 0.5, 0.6, and 0.2 respectively, then the other categories corresponding to the correlation coefficients of 0.5 and 0.6 are excluded, and the internal relationship between the key category and other categories is deeply explored to avoid excessive mutual influence between the correlation coefficients, thereby improving the accuracy of subsequent analysis.

[0108] Arrange the remaining categories according to the magnitude of the correlation coefficients with the degree of refrigerant leakage to generate a category importance ranking. Through this ranking, the influence degree of each category on the degree of refrigerant leakage can be intuitively displayed, providing an important reference basis for subsequent further analysis.

[0109] Select the top N groups of categories, and the steps to calculate the comprehensive analysis value are as follows;

[0110] Select the correlation coefficients between the top N groups of categories and the degree of refrigerant leakage and the correlation coefficients between the categories;

[0111] Use the correlation coefficient to calculate the evaluation value PGZ, and then calculate the adjustment coefficient PGs based on the evaluation value. The specific formula is as follows:

[0112]

[0113] In the formula, PGs iis the adjustment coefficient for the i-th category, PGZ i is the evaluation value for the i-th category, Qtxsi j is the correlation coefficient between the i-th category and the j-th category, QZ1 and QZ2 are weight coefficients, XGxsi is the correlation coefficient between the i-th category and the degree of refrigerant leakage, A is a constant coefficient, and 0.8 ≤ A ≤ 1;

[0114] This step is for further analysis and processing of the data, which can assign more accurate adjustment coefficients to each category, making subsequent operations more accurate and having a better usage effect.

[0115] Use the adjustment coefficient PGs to adjust the category data, and then superimpose the adjusted category data to obtain a comprehensive analysis value.

[0116] For example, if there are 3 groups of adjustment coefficients PGs, namely PGs1, PGs2, and PGs3, and the data for 3 categories corresponding to a certain time are SJ1, SJ2, and SJ3 respectively, then the comprehensive analysis value = PGs1 × SJ1 + PGs2 × SJ2 + PGs3 × SJ3.

[0117] Process multiple categories of data to avoid false prediction situations caused by sensor detection damage or incorrect operation, making the prediction accuracy higher and the usage effect better.

[0118] The present invention provides a big data-based automotive air-conditioning refrigerant slow-leakage fault prediction system and method, which uses big data technology to accurately collect historical data on refrigerant leakage of the target automotive air-conditioning model, processes the historical data of the target automotive air-conditioning model, extracts feature categories that can show refrigerant leakage, and further processes the feature categories to obtain a comprehensive analysis value that can significantly determine whether the refrigerant leaks, thereby facilitating subsequent accurate and rapid implementation of the fault prediction of automotive air-conditioning refrigerant slow leakage, having a good usage effect and a good application prospect.

[0119] 2.3. Data collection

[0120] During use, it is necessary to collect the working data of the automotive air-conditioning. Only by analyzing the working data of the automotive air-conditioning can the analysis and prediction of whether the automotive air-conditioning refrigerant leaks be realized. Therefore, it is necessary to collect the working data of the automotive air-conditioning and the corresponding feedback data, and this process is based on the collection module.

[0121] Collection module: Obtain automotive air-conditioning data in real time and calculate the real-time comprehensive analysis value according to the calculation logic of the feature module;

[0122] Obtaining automotive air-conditioning data in real time is based on the data collection function of the vehicle, directly obtaining the collected data and then performing operations.

[0123] Based on real-time comprehensive analysis value analysis, stable prediction of faults can be achieved with relatively high accuracy.

[0124] 2.4. Prediction Processing

[0125] In order to improve the accuracy of refrigerant prediction, it is necessary to further analyze and process the data. The method with high accuracy is to input the relevant data into the corresponding model. Therefore, it is necessary to analyze in combination with the model, and the analysis steps are based on the prediction module.

[0126] Prediction module: Obtain the operating data of the automotive air conditioner, input the air conditioner operating data and the real-time comprehensive analysis value into a pre-trained convolutional neural network model, obtain the quantitative evaluation percentage of cognitive or neural function, and obtain the refrigerant leakage percentage data for evaluating the degree of refrigerant leakage based on the cognitive or neural function;

[0127] The training method of the convolutional neural network model is as follows:

[0128] Divide the historical data set into a leakage degree training set and a leakage degree test set; the leakage degree training data includes the characteristic data reflecting the leakage degree and its corresponding quantitative score, and the characteristic data reflecting the leakage degree includes the refrigerant leakage percentage and the comprehensive analysis value;

[0129] Construct a convolutional neural network for performing regression tasks, use the characteristic data reflecting the leakage degree in the leakage degree training set as the input of the convolutional neural network, and use the refrigerant leakage percentage of the characteristic data reflecting the leakage degree as the output of the convolutional neural network, and train the convolutional neural network to obtain an initial convolutional neural network;

[0130] Use the leakage degree test set to verify the initial convolutional neural network model, and output the initial convolutional neural network with a test accuracy greater than or equal to the preset test accuracy as the trained convolutional neural network model.

[0131] Compare the refrigerant leakage percentage data with the leakage determination percentage threshold. If the refrigerant leakage percentage data is less than the leakage determination percentage threshold, it is determined that there is no slow leakage of the automotive air conditioner refrigerant, and no processing is performed; if the refrigerant leakage percentage data is not less than the leakage determination percentage threshold, it is determined that there is a slow leakage of the automotive air conditioner refrigerant. At this time, calculate the leakage rate based on the change of the leakage degree over time. The specific steps are as follows:

[0132] Collect K groups of consecutive refrigerant leakage percentage data, and calculate the difference data of the refrigerant leakage percentage between adjacent collection periods;

[0133] 3 < K < 6. For example, if K takes 4 groups, then there are 3 groups of difference data of the refrigerant leakage percentage between the calculated collection periods. Solve the first-order difference data of the refrigerant leakage percentage between adjacent collection periods to clearly present the change amount of refrigerant leakage within different time intervals.

[0134] On a continuous time series, sample K groups of data of the refrigerant leakage percentage. This data processing method can effectively capture the dynamic change data of refrigerant leakage within a specific period, and then provide key data support for in-depth analysis of the change trend of the refrigerant, accurately reflecting the actual situation of refrigerant leakage in a quantitative form.

[0135] Perform statistical analysis on K - 1 groups of difference data to obtain the average value Cp, and obtain the maximum value Cmax and the minimum value Cmin of the K - 1 groups of difference data.

[0136] Calculate the difference ratios between Cmax and Cmin and Cp respectively, that is, (Cmax - Cp) / Cp and (Cp - Cmin) / Cp, and compare the two groups of difference ratios with a preset fluctuation threshold. If both groups of difference ratios are less than the fluctuation threshold, then based on the stability and consistency of the data, the leakage rate can be determined as the average value Cp. If there is any group of difference ratios not less than the fluctuation threshold, given that the data fluctuation exceeds the expected range, indicating that there are significant changes in the leakage situation, then the leakage rate is determined as the maximum value Cmax.

[0137] This leakage data is the periodic leakage rate.

[0138] 2.5. Warning Alarm

[0139] When there is refrigerant leakage, corresponding warnings are needed to help the driver better formulate corresponding strategies. Therefore, after leakage, it is necessary to further analyze the leakage situation, and the further analysis of the leakage situation depends on the alarm module.

[0140] Alarm module: Obtain the refrigerant leakage degree data, calculate the leakage rate based on the change of the leakage degree over time, and then estimate the service life and risk value in turn. Then compare the service life and risk value with the preset rules and execute the corresponding warning strategy.

[0141] The steps to estimate the service life are as follows:

[0142] Obtain the refrigerant leakage percentage and leakage rate of the current state;

[0143] The refrigerant leakage percentage refers to the percentage of the refrigerant amount that has leaked in the system at the current time point relative to the initial refrigerant charge amount. The leakage rate represents the amount of refrigerant leakage per unit time. These two parameters are the basic data for subsequent estimation work, and their accuracy directly affects the reliability of the estimation results.

[0144] Compare the leakage rate with the preset leakage rate threshold levels, and adjust the leakage rate based on the correction factor corresponding to the leakage rate threshold level.

[0145] The leakage rate threshold levels are a series of critical values determined comprehensively based on various factors such as the type, specifications, and safe operation standards of the refrigeration system. Each threshold level corresponds to a different risk level and correction factor. When the leakage rate is within a certain threshold level range, the original leakage rate is adjusted based on the correction factor corresponding to that level to more accurately reflect the impact of the actual refrigerant leakage trend of the system on its service life. For example, if the leakage rate is higher than a certain higher threshold level, the corresponding correction factor will increase the adjusted leakage rate, indicating that the negative impact of refrigerant leakage on the system life is more serious.

[0146] For example, if the leakage rate is the maximum value Cmax, and the correction factor corresponding to its leakage rate threshold level is A3, then the adjusted leakage rate is Cmax × A3. If A3 is greater than 1 at this time, the adjusted leakage rate is greater than the leakage rate determined in the previous step.

[0147] Calculate the difference between the critical refrigerant usage percentage and the refrigerant leakage percentage, and use this difference and the adjusted leakage rate to estimate the service life.

[0148] For example, if the difference is positive and the adjusted leakage rate is known, the estimated remaining service time can be obtained by dividing the difference by the leakage rate. This time represents the time remaining for the system to reach the critical refrigerant usage state from the current state under the current leakage trend, thereby realizing the estimation of the refrigerant service life.

[0149] For example, the calculated difference between the critical refrigerant usage percentage and the refrigerant leakage percentage is B1, and the adjusted leakage rate is Cmax × A3. Then the calculated refrigerant service life Ts = B1 / (Cmax × A3) × T1 - Q1, where Q1 is the preset time for ensuring safety and reaction duration, and T1 is the cycle duration, thereby obtaining the refrigerant service life.

[0150] The steps to estimate the risk value are as follows:

[0151] Obtain the refrigerant leakage percentage, service life, and the leakage rate threshold level corresponding to the leakage rate in the current state.

[0152] The leakage rate threshold levels corresponding to the refrigerant leakage percentage, service life, and leakage rate are processed using a weighted summation calculation method to obtain a risk value. The formula for the risk value is as follows:

[0153] FXsz = XL × W1 + (Tb - Ts) × W2 + Dj 2 × W3

[0154] In the formula, FXsz is the risk value, XL is the refrigerant leakage percentage in the current state, Tb is the preset upper limit service duration based on the historical leakage data of the air conditioner, Ts is the service life, Dj is the leakage rate threshold level, and W1, W2, and W3 are the weight ratios.

[0155] Compare the service life and the risk value with the preset rules, and the steps to execute the corresponding warning strategy are as follows:

[0156] Judge whether the service life is within the life danger range and whether the risk value is within the leakage danger range;

[0157] If the service life is within the life danger range and the risk value is within the leakage danger range, it is determined that the leakage is serious, a warning is issued, the car air conditioner is turned off, and a parking instruction is sent;

[0158] If only the service life is within the life danger range or the risk value is within the leakage danger range, it is determined that the leakage is relatively serious, a warning is issued, the car air conditioner is turned off, and the nearest maintenance location is pushed;

[0159] If there is no situation where the service life is within the life danger range and the risk value is within the leakage danger range, it is determined that the leakage is relatively minor, a warning is issued, high-intensity use of the car air conditioner is prohibited, and the maintenance location is pushed;

[0160] The warning includes, but is not limited to, timely transmitting information about the abnormal state of the equipment or asset to relevant personnel through system pop-ups, SMS pushes, email notifications, etc., to ensure that effective countermeasures can be taken in a timely manner to reduce potential risks.

[0161] The present invention provides a system and method for predicting slow refrigerant leakage faults in car air conditioners based on big data. It further analyzes on the basis of showing the characteristic categories of refrigerant leakage. By analyzing and mining the complex data relationship between the comprehensive analysis value and cognitive or neural functions, it provides a more accurate and comprehensive tool for evaluating the degree of refrigerant leakage. Compared with traditional methods, the present invention can accurately achieve the degree of refrigerant leakage, and further predict the service life and the risk value of use, laying a foundation for users to make reasonable choices, thereby providing a more suitable usage plan to help users better use the car air conditioner, and having good application prospects.

[0162] The weight coefficient is determined by the coefficient of variation method. The coefficient of variation method is a method of assigning weights to each evaluation index according to the degree of variation between the current value and the target value of each evaluation index. If the numerical difference of a certain index is large and can clearly distinguish each evaluated object, it means that the discrimination information of this index is rich, so a larger weight should be given to this index. On the contrary, if the numerical differences of each evaluated object on a certain index are small, then the ability of this index to distinguish each evaluation object is weak, so a smaller weight should be given to this index. This method directly uses the information contained in each index and calculates the weight of the index, so it has objectivity.

[0163] Embodiment 2

[0164] A method for predicting slow refrigerant leakage faults of automotive air conditioners based on big data, as Figure 2 shown, includes the following steps:

[0165] Use big data technology to screen and refine the historical data of refrigerant leakage of the target automotive air conditioner model from the database and perform preprocessing;

[0166] Apply a feature selection algorithm to analyze the preprocessed data, generate a ranking of category importance, select the top N groups of categories, calculate the comprehensive analysis value, and summarize and construct a historical data set;

[0167] Obtain automotive air conditioner data in real time and calculate the real-time comprehensive analysis value;

[0168] Obtain the operating data of the automotive air conditioner, input the air conditioner operating data and the real-time comprehensive analysis value into a pre-trained convolutional neural network model, obtain the quantitative evaluation percentage of cognitive or neurological function, and obtain the refrigerant leakage percentage data for evaluating the degree of refrigerant leakage according to the cognitive or neurological function;

[0169] Obtain the data of the degree of refrigerant leakage, calculate the leakage rate based on the change of the leakage degree over time, and sequentially estimate the service life and the risk value, and then compare the service life and the risk value with the preset rules to execute the corresponding warning strategy.

[0170] In the application, several formulas involved are calculated by taking their numerical values after dimensionless, and the establishment of the formulas is obtained by collecting a large amount of data for software simulation to get a formula closest to the real situation. Some coefficients or weights in the formulas are set by those skilled in the art according to the actual situation, so no more details will be elaborated here.

[0171] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any combination thereof; when implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product; those of ordinary skill in the art will appreciate that the units and algorithm steps of the examples described in connection with the embodiments disclosed herein can be implemented in electronic hardware or in a combination of computer software and electronic hardware; whether these functions are executed in hardware or software depends on the specific application and design constraints of the technical solution.

[0172] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units. They may be located in one place or distributed to multiple network units; some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0173] As described above, the specific implementation manners of the present application are only described, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed in the present application, and all of them should be covered by the protection scope of the present application.

Claims

1. An automotive air-conditioning refrigerant slow-leakage fault prediction system based on big data, characterized in that, Including: Sorting module: Using big data technology to screen and refine the historical data of refrigerant leakage of the target automotive air conditioner model from the database and perform preprocessing; Feature module: Analyze the preprocessed data using a feature selection algorithm to generate a category importance ranking, select the top N groups of categories, calculate the comprehensive analysis value, and summarize and construct a historical dataset; Collection module: Real-time obtain automotive air conditioner data and calculate the real-time comprehensive analysis value according to the calculation logic of the feature module; Prediction module: Obtain the operating data of the automotive air conditioner, input the air conditioner operating data and the real-time comprehensive analysis value into a pre-trained convolutional neural network model to obtain the quantitative evaluation percentage of cognitive or neurological function, and obtain the refrigerant leakage percentage data for evaluating the refrigerant leakage degree based on the cognitive or neurological function; Alarm module: Obtain the refrigerant leakage degree data, calculate the leakage rate based on the change of the leakage degree over time, and then estimate the service life and risk value in sequence. Then compare the service life and risk value with the preset rules and execute the corresponding warning strategy.

2. The vehicle air conditioner refrigerant slow leakage fault prediction system based on big data according to claim 1, wherein: The historical data of refrigerant leakage of the automotive air conditioner includes but is not limited to the automotive environmental temperature, condenser inlet pressure, condenser inlet temperature, condenser outlet pressure, condenser outlet temperature, compressor suction temperature, compressor discharge temperature, compressor suction pressure, compressor discharge pressure, compressor power, evaporator inlet pressure, evaporator inlet temperature, evaporator outlet pressure, evaporator outlet temperature, and refrigerant pipeline vibration parameters when the refrigerant leaks.

3. The automotive air-conditioning refrigerant slow-leakage fault prediction system based on big data according to claim 2, characterized in that: The steps for preprocessing the historical data of refrigerant leakage are as follows: Compare the identification codes of the data and remove duplicate data; Use the mean filling method to process missing values; Integrate the processed data, and the data of each category forms a complete dataset.

4. A big data-based automotive air conditioner refrigerant slow leakage fault prediction system according to claim 3, characterized in that: The steps for analyzing the preprocessed data using a feature selection algorithm to generate a category importance ranking are as follows: Use the Z-score standardization method to standardize the data in the dataset, and then use the min-max normalization method to normalize the data; Analyze the data using the Pearson correlation coefficient method and calculate the correlation coefficients between all categories and the refrigerant leakage degree; Obtain the category with the largest absolute value of the correlation coefficient with the refrigerant leakage degree, mark it as the key category, and then calculate the correlation coefficients between the key category and other categories in sequence; Compare the correlation coefficients between the key category and other categories with the preset standard interval and delete the categories not within the standard interval; Arrange the remaining categories according to the magnitude of the correlation coefficients with the refrigerant leakage degree to generate a category importance ranking.

5. The automotive air-conditioning refrigerant slow-leakage fault prediction system based on big data according to claim 4, wherein: The steps for selecting the top N groups of categories and calculating the comprehensive analysis value are as follows; Select the correlation coefficients between the top N groups of categories and the refrigerant leakage degree and the correlation coefficients between the categories; Calculate the evaluation value PGZ using the correlation coefficient, and then calculate the adjustment coefficient PGs based on the evaluation value. The specific formulas are as follows: wherein, PGs i is the adjustment coefficient of the i-th category, PGZ i is the evaluation value of the i-th category, Qtxsi j is the correlation coefficient between the i-th category and the j-th category, QZ1 and QZ2 are weight coefficients, XGxsi is the correlation coefficient between the i-th category and the degree of refrigerant leakage, A is a constant coefficient, 0.8 ≤ A ≤ 1; Use the adjustment coefficient PGs to adjust the category data, and then superimpose the adjusted category data to obtain the comprehensive analysis value.

6. The vehicle air conditioner refrigerant slow leakage fault prediction system based on big data according to claim 5, characterized in that: The training method of the convolutional neural network model is as follows: Divide the historical dataset into a leakage degree training set and a leakage degree test set; the leakage degree training data includes leakage degree reflection feature data and its corresponding quantization score, and the leakage degree reflection feature data includes the refrigerant leakage percentage and the comprehensive analysis value; Construct a convolutional neural network for performing regression tasks, use the leakage degree reflection feature data in the leakage degree training set as the input of the convolutional neural network, and use the refrigerant leakage percentage of the leakage degree reflection feature data as the output of the convolutional neural network, and train the convolutional neural network to obtain an initial convolutional neural network; Use the leakage degree test set to verify the model of the initial convolutional neural network, and output the initial convolutional neural network with a preset test accuracy greater than or equal to as the trained convolutional neural network model.

7. A big data-based automotive air conditioner refrigerant slow leakage fault prediction system according to claim 6, characterized in that: Compare the refrigerant leakage percentage data with the leakage determination percentage threshold. If the refrigerant leakage percentage data is less than the leakage determination percentage threshold, it is determined that there is no slow leakage of the automotive air-conditioning refrigerant, and no treatment is required; if the refrigerant leakage percentage data is not less than the leakage determination percentage threshold, it is determined that there is a slow leakage of the automotive air-conditioning refrigerant. At this time, calculate the leakage rate based on the change of the leakage degree over time. The specific steps are as follows: Collect K groups of refrigerant leakage percentage data continuously, and calculate the difference data of the refrigerant leakage percentage between adjacent collection periods; Perform statistical analysis on the K-1 groups of difference data to obtain the average value Cp, and obtain the maximum value Cmax and the minimum value Cmin of the K-1 groups of difference data; Calculate the difference ratios between Cmax and Cmin and Cp respectively, and compare the two groups of difference ratios with the preset fluctuation threshold. If both groups of difference ratios are less than the fluctuation threshold, it is determined that the leakage rate is the average value Cp. If there is any group of difference ratios not less than the fluctuation threshold, it is determined that the leakage rate is the maximum value Cmax.

8. The automotive air-conditioning refrigerant slow-leakage fault prediction system based on big data according to claim 7, characterized in that: The steps for estimating the service life are as follows: Obtain the refrigerant leakage percentage and leakage rate of the current state; Compare the leakage rate with the preset leakage rate threshold level, and adjust the leakage rate based on the correction coefficient corresponding to the leakage rate threshold level; Calculate the difference between the critical refrigerant usage percentage and the refrigerant leakage percentage, and use this difference and the adjusted leakage rate to estimate the service life.

9. The system for predicting the slow refrigerant leakage fault of an automotive air conditioner based on big data according to claim 8, wherein: The steps for estimating the risk value are as follows: Obtain the refrigerant leakage percentage, service life, and leakage rate threshold level corresponding to the leakage rate of the current state; Use the calculation method of weighted summation to process the refrigerant leakage percentage, service life, and leakage rate threshold level corresponding to the leakage rate to obtain the risk value.

10. A method for predicting the slow refrigerant leakage fault of an automotive air conditioner based on big data, using the system described in any one of claims 1 to 9, characterized in that: Include the following steps: Use big data technology to screen and refine the historical data of refrigerant leakage of the target automotive air-conditioning model from the database, and perform preprocessing; Use the feature selection algorithm to analyze the preprocessed data, generate a category importance ranking, select the first N groups of categories before the ranking, calculate the comprehensive analysis value, and summarize and construct a historical dataset; Obtain the automotive air-conditioning data in real time, and calculate the real-time comprehensive analysis value; Obtain the operating data of the automotive air conditioner, input the air conditioner operating data and the real-time comprehensive analysis value into a pre-trained convolutional neural network model, obtain the quantitative evaluation percentage of cognitive or neurological functions, and obtain the refrigerant leakage percentage data for evaluating the degree of refrigerant leakage based on the cognitive or neurological functions; Obtain the refrigerant leakage degree data, calculate the leakage rate based on the change of the leakage degree over time, and sequentially estimate the service life and the risk value. Then compare the service life and the risk value with the preset rules and execute the corresponding warning strategy.

Images