A production management system and management method for new energy automobile air conditioning components

Abstract

The application discloses a new energy automobile air conditioner component production management system and management method, and relates to the technical field of automobile air conditioner production; comprising a data acquisition module, a defect analysis module, an abnormality analysis module, a comprehensive analysis module and a normality processing module; the data acquisition module is used for collecting historical state and real-time new energy automobile air conditioner component data, and the technical key points are as follows: the new energy automobile air conditioner component historical detection data can be collected, and the historical data can be further analyzed to research standard audio features and standard intervals of standard new energy automobile air conditioner components, so that the processed real-time collected new energy automobile air conditioner component historical detection data can be compared with the standard audio features and the standard intervals in the subsequent process, whether defects exist can be quickly judged, relevant technical personnel can conveniently understand the situation of the produced new energy automobile air conditioner components, the use is relatively convenient, and the application has a good use prospect.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of automobile air conditioner production, in particular to a production management system and management method for new energy automobile air conditioner components. BACKGROUND

[0002] With the rapid growth of the automobile industry in the past few years, especially passenger cars, the automobile parts industry has also developed rapidly. Automobile air conditioners, as an important component for improving the comfort of car riding, have been recognized by automobile manufacturers and consumers. The installation rate of domestic passenger car air conditioners is close to 100%, and the installation rate on other vehicle models is also increasing year by year. Automobile air conditioning devices have become a functional component in automobiles.

[0003] The automobile air conditioning device is referred to as an automobile air conditioner, which is used to adjust and control the temperature, humidity, air cleanliness and air flow in the vehicle cabin to the best state, provide a comfortable riding environment for passengers, reduce travel fatigue, and create good working conditions for drivers, which plays an important role in ensuring safe driving.

[0004] The automobile air conditioner generally includes a refrigeration device, a heating device and a ventilation device. This combined device makes full use of the limited space inside the automobile and has a simple structure and is easy to operate.

[0005] The new energy automobile air conditioner is mainly composed of a pure electric or hybrid compressor, a condenser, a liquid accumulator, an expansion valve, an evaporative box and a control circuit. The low-pressure pipeline: from the outlet of the throttle valve to the inlet of the compressor, along the way there are evaporative boxes, low-pressure filling ports and accumulators. The high-pressure pipeline: from the outlet of the compressor to the inlet of the throttle valve, along the way there are compressors, condensers, dryers, high-pressure filling ports, high-low pressure switches and throttle valves.

[0006] With the progress of society and the rapid development of industry, in order to better produce new energy automobile air conditioners, people have invented some production management systems for new energy automobile air conditioner components;

[0007] The existing patent No. CN111915134B patent entitled "A property intelligent community management system and management method" records that the automobile air conditioner compressor is made of various compressor parts, each compressor part is provided with a unique product code, and the production management system is provided with an input unit, a management storage unit, an output unit and a man-machine interaction unit; based on the production management method of the automobile air conditioner compressor, comprising the following steps: 1. Information input, 2. Data collection and analysis, 3. Workpiece progress display, 4. Man-machine interaction. The product code of the workpiece is recorded, the access status of the workpiece is recorded, the material verification and settlement are facilitated; the processing and assembling time, quality and quantity of the workpiece are recorded in real time, the production plan is arranged, the number of employees in different processes is adjusted, the work efficiency is maximized; the workpiece position of unqualified workpiece is rectified, the production problem is solved, the work efficiency and quality are improved;

[0008] The central idea of the scheme recorded in the above patent is to encode the workpiece of self-made, outsourcing and purchasing, effectively record the access status of the workpiece, facilitate material verification and settlement; real-time record and track the processing and assembling time, quality and quantity of the workpiece, facilitate overall arrangement of production plan, corresponding adjustment of number of employees in different processes, realization of maximization of work efficiency; targeted rectification of workpiece position of unqualified workpiece, efficient and rapid solution of production problem, improvement of work efficiency and quality; timely alarm and reminder of inventory products and slow-selling products, preferential sale of inventory products and slow-selling products, reduction of inventory quantity and increase of economic benefit; market procurement and planned production combined with MRP, adaptation to market demand and reduction of procurement expenditure.

[0009] However, in actual use, the above system plays a prompting role, which is equivalent to the arrangement and scheduling of existing generated data;

[0010] The above method plays a certain effect in the production process of air conditioner in comprehensive scene, which is based on the processed data result transmitted by the user or existing system in real time, but in the production process, the difficulty is not adjustment after finding the problem, but how to find the problem, especially the problem of new energy automobile air conditioner part production. At present, for new energy automobile air conditioner parts, as the automatic equipment is more and more, and most of the parts are processed once, therefore, the qualified rate of new energy automobile air conditioner parts is more than 95%, and after the parts are assembled and welded to form new energy automobile air conditioner parts, the qualified rate is usually reduced to about 89%, therefore, it is necessary to analyze the new energy automobile air conditioner parts and understand their defects, for this purpose, we have developed a production management system and management method for new energy automobile air conditioner parts. SUMMARY

[0011] (I) Technical problems solved

[0012] In view of the deficiencies of the prior art, the new energy automobile air conditioner component production management system and management method can collect historical detection data of new energy automobile air conditioner components, and can further analyze the historical data to research standard audio features and standard intervals of standard new energy automobile air conditioner components, so that the processed real-time collected historical detection data of new energy automobile air conditioner components can be compared with the standard audio features and the standard intervals, whether there is a defect can be quickly judged, relevant technical personnel can understand the situation of the produced new energy automobile air conditioner components, the use is convenient, the overall use effect is good, has good use prospect, and effectively solves the problems in the background art.

[0013] (II) Technical solutions

[0014] To achieve the above object, the following technical solutions are used:

[0015] A new energy automobile air conditioner component production management system, comprising a data acquisition module, a defect analysis module, an abnormality analysis module, a comprehensive analysis module and a normality processing module:

[0016] The data acquisition module is used to collect historical state and real-time new energy automobile air conditioner component data, the real-time new energy automobile air conditioner component data includes audio data of new energy automobile air conditioner component commissioning detection and picture data of new energy automobile air conditioner part assembly welding positions after detection, and the historical state new energy automobile air conditioner component data is audio data of new energy automobile air conditioner component commissioning detection without defects;

[0017] The defect analysis module is used to analyze the audio data of the historical state new energy automobile air conditioner component commissioning detection, set standard audio features and standard intervals, then process the real-time collected audio data of the new energy automobile air conditioner component commissioning detection, compare the processed audio features with the standard audio features, calculate the real-time difference ratio, and compare the real-time difference ratio with the standard interval to judge whether the new energy automobile air conditioner component has defects;

[0018] The abnormality analysis module is used to start when a defect signal is received, calculate the real-time defect abnormality level, predict the defect cause according to the audio performance, mark the new energy automobile air conditioner part assembly welding position with defects according to the defect cause, and summarize to form a predicted defect list;

[0019] The comprehensive analysis module is used to obtain picture data of the new energy automobile air conditioner part assembly welding position corresponding to the predicted defect list, and analyze the obtained picture data to judge whether there is a defect;

[0020] If there is a defect, the defect data is input into the database, and the defect anomaly level is predicted based on the defect data, the defect level difference between the predicted defect anomaly level and the real-time defect anomaly level is calculated, and the defect level difference is compared with the preset defect variation interval. If it is located inside the defect variation interval, the defect formation reason is found based on the defect data, and the defect is repaired;

[0021] If there is no defect or the defect level difference is located outside the defect variation interval, the new energy automobile air conditioner part assembly welding position on the predicted defect list is comprehensively detected, the defect area and the defect formation reason are obtained according to the comprehensive detection data, and the defect is repaired;

[0022] Normal processing module: for randomly extracting the defect area of W new energy automobile air conditioner parts assembly in this batch for detection, judging whether there is a defect, and calculating the defect rate;

[0023] If the defect rate is greater than the set value, the corresponding processing scheme is adjusted according to the defect reason;

[0024] If the defect rate is less than the set value, no processing is done.

[0025] Further, the audio data of the historical state new energy automobile air conditioner part test running detection is analyzed, including analyzing the main feature and the auxiliary spectrum feature, and the steps of analyzing the main feature are as follows:

[0026] The obtained audio data of the historical state new energy automobile air conditioner part test running detection is divided into N frames;

[0027] The divided N frames of audio data are converted into frequency domain representation by Fourier transformation;

[0028] The converted signal spectrum is filtered using a Mel filter;

[0029] The signal output by the Mel filter is logarithmically operated to convert it into a Mel frequency coefficient;

[0030] The Mel frequency coefficient is subjected to discrete cosine transformation to obtain a preliminary feature coefficient;

[0031] The preliminary feature coefficient is subjected to energy item removal and difference item processing to obtain an MFCC feature vector;

[0032] The auxiliary spectrum feature includes any A of the following: spectrum energy, spectrum centroid, spectrum entropy, spectrum peak, spectrum attenuation, spectrum flux, spectrum kurtosis, spectrum slope, spectrum skewness, spectrum spread, and spectrum roll-off point.

[0033] Further, the steps of setting the standard audio data are as follows:

[0034] Obtaining the main feature and auxiliary spectral feature data of the audio data of the historical state new energy automobile air conditioning component test run detection of the last K groups of calculation;

[0035] Calculating the average data of the K groups of main features and auxiliary spectral feature data, obtaining the average of the N MFCC feature vectors and the average of the A groups of auxiliary spectral features, and summarizing the average data to obtain the standard audio feature;

[0036] Respectively calculating the difference ratio of the K groups of data and the average data, obtaining the maximum and minimum of the single feature difference ratio, and the maximum and minimum of the comprehensive feature difference ratio, and obtaining the single standard interval and the comprehensive standard interval;

[0037] The formula for calculating the single feature difference ratio is as follows:

[0038]

[0039] In the formula, Diffd is the single feature difference ratio, Tz is the audio feature data, is the average of the K groups of audio feature data;

[0040] The calculation formula of the comprehensive feature difference ratio is as follows:

[0041]

[0042] In the formula, Diffz is the comprehensive feature difference, Tzz i is the i-th MFCC feature vector data, is the average calculated based on all i-th MFCC feature vector data of the K groups of historical data, Xsz is the weight coefficient of the MFCC feature vector data difference ratio, Tzf x is the x-th auxiliary spectral feature data, is the average of all x-th auxiliary spectral feature data of the K groups of historical data, Xsf x is the weight coefficient of the x-th auxiliary spectral feature data difference ratio.

[0043] Further, the real-time difference ratio includes real-time single difference ratio and real-time comprehensive difference ratio, and the comparison between the real-time difference ratio and the standard interval includes the comparison between the real-time single difference ratio and the single standard interval, and the comparison between the real-time comprehensive difference ratio and the comprehensive standard interval;

[0044] When judging whether the new energy automobile air conditioning component has defects;

[0045] If the real-time single difference is located in the single standard interval, and the real-time comprehensive difference ratio is located in the comprehensive standard interval, it is determined that there is no defect;

[0046] All other cases are defective.

[0047] Further, the formula for calculating the real-time defect anomaly level is as follows:

[0048]

[0049] In the formula, Def is the calculated real-time defect anomaly level, Def=ROUNDUP() is the upward rounding function, Defd is the distance between the real-time single difference value proportion and the single standard interval, Blxsd is the proportion coefficient of the distance between the real-time single difference value proportion and the single standard interval, Defz is the distance between the real-time comprehensive difference value proportion and the comprehensive standard interval, Blxsz is the proportion coefficient of the distance between the real-time comprehensive difference value proportion and the comprehensive standard interval, B is the distance difference corresponding to the preset corresponding anomaly level, L is the number of defects existing in the N MFCC feature vectors, cz i is the real-time single difference value proportion of the i-th defect MFCC feature vector, Maxz i is the maximum difference value proportion of the i-th defect MFCC feature vector, Minz i is the minimum difference value proportion of the i-th defect MFCC feature vector, Defd1 is the distance between the auxiliary spectral feature data and the corresponding single standard interval, Z is the number of defects existing in the auxiliary spectral feature data, czf j is the real-time single difference value proportion of the j-th defect auxiliary spectral feature data, Maxf j is the maximum difference value proportion of the j-th defect auxiliary spectral feature data, Minf j is the minimum difference value proportion of the j-th defect auxiliary spectral feature data, Xsf j is the weight coefficient of the difference value proportion of the j-th defect auxiliary spectral feature data, Zcz is the real-time comprehensive difference value proportion, Zmax is the maximum value of the comprehensive standard interval, and Zmin is the minimum value of the comprehensive standard interval.

[0050] Further, when predicting the defect causes based on the audio performance, all defect feature data is input into the database, 2-3 groups of historical abnormal data closest to each defect feature data are found in the database, and the defect causes recorded in the defect abnormal data are called.

[0051] Further, the steps of analyzing the obtained picture data to determine whether there is a defect are as follows:

[0052] The obtained picture data is called, and the picture data is processed by using a threshold segmentation method to remove the picture background;

[0053] Gaussian filtering is used to denoise the picture after removing the background, and then grayscale processing is performed;

[0054] The Sauvola threshold method is used to binarize the pixel points of the gray-scale processed picture, and the defect image on the picture is segmented.

[0055] The segmented image is a defect image.

[0056] When the defect image exists, the number of pixel points in the segmented image and the gray value of the pixel points in the image are obtained.

[0057] Further, the formula for predicting the defect anomaly level based on the defect data is as follows:

[0058]

[0059] In the formula, Def1 is the predicted defect anomaly level, D is the number of defect images, hdc i is the gray data of the i-th defect image, hsxs i is the conversion ratio, E is the total number of pixel points of the i-th defect image, hd j is the gray value of the j-th pixel point in the i-th defect image, hdb j is the standard value set for the i-th defect image, and bdz is the difference value, 0 < bdz < 5.

[0060] The defect level difference Defc = Def-Def1, and the preset defect variation range is where Def max is the set maximum defect anomaly level, and β is a constant data, 25 ≤ β ≤ 35.

[0061] Further, when detecting the defect area, the detection range is centered on the defect area, and the detection area is γ times the defect area, γ > 3.

[0062] The calculated defect rate S q is the number of parts with defects at that position in W new energy vehicle air conditioning parts, and the set value is δ, 3% < δ < 5%.

[0063] Further, a production management method for new energy vehicle air conditioning components includes the following steps:

[0064] Collect historical state and real-time new energy vehicle air conditioning component data, real-time new energy vehicle air conditioning component data includes audio data of new energy vehicle air conditioning component test run detection and picture data of new energy vehicle air conditioning part assembly welding position after detection, and historical state new energy vehicle air conditioning component data is audio data of new energy vehicle air conditioning component test run detection without defects.

[0065] The audio data of the historical state new energy automobile air conditioning component commissioning detection is analyzed, the standard audio features and the standard interval are set, then the audio data of the real-time collected new energy automobile air conditioning component commissioning detection is processed, the processed audio features are compared with the standard audio features, the real-time difference ratio is calculated, the real-time difference ratio is compared with the standard interval, whether the new energy automobile air conditioning component has defects is judged;

[0066] When the defect existing signal is received, the real-time defect abnormality level is calculated, the defect reason is predicted according to the audio performance, the new energy automobile air conditioning part assembly welding position existing defects is marked according to the defect reason, and the predicted defect list is formed by summarizing;

[0067] The picture data of the new energy automobile air conditioning part assembly welding position corresponding to the predicted defect list is obtained, and the obtained picture data is analyzed to determine whether there is a defect;

[0068] If there is a defect, the defect data is input into the database, the defect abnormality level is predicted based on the defect data, the defect level difference between the predicted defect abnormality level and the real-time defect abnormality level is calculated, and the defect level difference is compared with the preset defect variation interval, if it is located in the defect variation interval, the defect forming reason is found based on the defect data, and the defect is repaired;

[0069] If there is no defect or the defect level difference is located outside the defect variation interval, the new energy automobile air conditioning part assembly welding position on the predicted defect list is comprehensively detected, the defect area and the defect forming reason are obtained according to the comprehensive detection data, and the defect is repaired;

[0070] Randomly extract the defect area of the batch W new energy automobile air conditioning part assembly for detection, judge whether there is a defect, and calculate the defect rate;

[0071] If the defect rate is greater than the set value, the corresponding processing scheme is adjusted according to the defect reason;

[0072] If the defect rate is less than the set value, no processing is performed.

[0073] (Three) beneficial effects

[0074] The present application provides a new energy automobile air conditioning component production management system and management method, which has the following beneficial effects:

[0075] 1. The application provides a new energy vehicle air conditioner component production management system and management method, which can collect new energy vehicle air conditioner component historical detection data, and can further analyze the historical data to study standard audio features and standard intervals of standard new energy vehicle air conditioner components, so as to facilitate subsequent comparison of processed real-time collected new energy vehicle air conditioner component historical detection data with standard audio features and standard intervals, can quickly determine whether there is a defect, can facilitate relevant technical personnel to understand the situation of the produced new energy vehicle air conditioner components, is convenient to use, has good overall use effect, and has good use prospect.

[0076] 2. The application provides a new energy vehicle air conditioner component production management system and management method, which can quickly analyze and judge the assembled new energy vehicle air conditioner components when in use, judge whether there is a defect, and can further analyze and predict the defect cause, and can automatically call the picture corresponding to the defect cause, further analyze the picture, and judge the defect formation cause. This kind of sequential judgment can reduce the analysis amount of the picture, can effectively reduce the data processing amount, and is a comprehensive detection, can reflect the cooperation between parts and the welding effect, further reflects the assembly capacity of the enterprise for parts, has good use effect, and has good use prospect.

[0077] 3. The application provides a new energy vehicle air conditioner component production management system and management method, which analyzes and judges the new energy vehicle air conditioner components, can analyze the defect formation cause and the position of the new energy vehicle air conditioner components, and detects all new energy vehicle air conditioner components of the same batch in a targeted manner, can help relevant technical personnel to understand whether the new energy vehicle air conditioner components belong to normal problems, so as to facilitate relevant management personnel to formulate corresponding production strategies, so that the qualified rate of the produced new energy vehicle air conditioner components is higher, has good use effect, and has good application scenarios. BRIEF DESCRIPTION OF DRAWINGS

[0078] Figure 1 The application provides a flowchart of a new energy vehicle air conditioner component production management system. DETAILED DESCRIPTION

[0079] The technical solutions in the embodiments of the application will be described clearly and completely in combination with the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, not all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the application.

[0080] The central idea of the existing scheme is to encode the workpieces of self-made, outsourcing and purchased products, effectively record the access status of the workpieces, facilitate material verification and settlement; real-time record and track the processing and assembly time, quality and quantity of the workpieces, facilitate overall arrangement of production plan, adjust the number of employees in different processes accordingly, maximize work efficiency; targeted rectification of workstations with unqualified workpieces, facilitate efficient and rapid solution of production problems, improve work efficiency and quality; timely alarm for inventory products and slow-selling products, preferentially sell inventory products and slow-selling products, reduce inventory, increase economic benefits; market procurement and planned production combined with MRP, adapt to market demand, reduce procurement expenses.

[0081] However, in actual use, the number of automated equipment gradually increases, so the pass rate of new energy automobile air conditioner parts production is gradually increasing, and the role of defect management is gradually reduced, so that the new energy automobile air conditioner part production management system plays a prompting role, which is equivalent to the arrangement and mobilization of existing data generated, and the overall effect is very limited.

[0082] Specifically, the above-mentioned method plays a certain role in the production process of air conditioners in a comprehensive scene, and is based on the processed data results transmitted in real time by users or existing systems, and the arrangement of market feedback data, and its role is to arrange the analyzed result data.

[0083] However, in the production process, the difficulty is not in the arrangement of data, but in how to find problems, especially in the production of new energy automobile air conditioner parts. At present, for new energy automobile air conditioner parts, as the number of automated equipment increases, and most of the parts are processed once (one-time forming), the pass rate of new energy automobile air conditioner parts is more than 95%, and after the parts are assembled and welded to form new energy automobile air conditioner parts, the pass rate is usually reduced to about 89% (welding and assembly require multiple equipment cooperation, and when multiple equipment cooperate, the cooperation between equipment and the error of equipment itself will be magnified, so the pass rate will be reduced), therefore, it is necessary to analyze the production of new energy automobile air conditioner parts and understand its defects, for this purpose, we have developed a new energy automobile air conditioner part production management system and management method.

[0084] Therefore, in the development process, the initial development idea is how to realize the detection of new energy automobile air conditioner parts, and judge whether the new energy automobile air conditioner parts are damaged, the initial idea is to take pictures of the assembled new energy automobile air conditioner parts, and comprehensively analyze the pictures taken, so as to realize the detection of the whole new energy automobile air conditioner parts, which can realize rapid and comprehensive detection of large defects.

[0085] However, in actual use, it is found that the direct shooting of new energy vehicle air conditioning components for detection does not expose many problems that have not been exposed through actual operation, so the new energy vehicle air conditioning components photographed are basically shown to be defect-free, and in this case, the desired defect data cannot be obtained, so the use effect is not good.

[0086] Therefore, in the middle of the research and development, the defect detection scheme is re-studied, and later it is found that in order to ensure that the automobile air conditioning components can be used, the operation detection of each component is still usually carried out at the time of leaving the factory, so the data acquisition method is changed, and the time node of acquiring data is moved to this time point, so that the detection of new energy vehicle air conditioning components can be realized, and the detection effect is better.

[0087] However, in the process of actual use, it is found that direct detection usually obtains multiple defect reasons, and further detection is still needed to understand the specific reasons for the formation of defects, and then the corresponding repair scheme and production strategy can be formulated according to the defect reasons, so the above scheme has certain defects in use and needs to be further improved.

[0088] Therefore, in the later stage of research and development, a new energy vehicle air conditioning component production management system is researched and developed, which can help relevant personnel better understand the defects of new energy vehicle air conditioning components and help analyze whether the defects of new energy vehicle air conditioning components belong to normal problems.

[0089] Embodiment 1

[0090] The hardware part of the new energy vehicle air conditioning component production management system is based on the server of the existing production system, and the data collected is directly obtained from the detection equipment connected with the Internet of Things, so it is not necessary to separately increase new data acquisition devices, and therefore the cost of the whole system application is relatively low.

[0091] The software part of the new energy vehicle air conditioning component production management system includes a data acquisition module, a defect analysis module, an abnormality analysis module, a comprehensive analysis module and a normality processing module, and the specific process is shown in Figure 1 :

[0092] The data required for the operation of the system is the detection data of new energy vehicle air conditioning components (new energy vehicle air conditioning components are new energy vehicle air conditioning parts assembled and welded), and the data acquisition process is based on the data acquisition module.

[0093] The data acquisition module is used to collect historical state and real-time new energy vehicle air conditioning component data, the real-time new energy vehicle air conditioning component data including audio data of new energy vehicle air conditioning component test run detection and picture data of new energy vehicle air conditioning part assembly welding position after detection, and the historical state new energy vehicle air conditioning component data being audio data of new energy vehicle air conditioning component test run detection without defects;

[0094] The new energy vehicle air conditioning component detection is to test the new energy vehicle air conditioning component under the condition of 2-5 times of the limit working pressure for 20 minutes (normal defects are exposed after running for this time, and normal detection methods are difficult to detect, and the test time can be appropriately prolonged), and the last 3 minutes of the test run detection audio data is taken as the audio data of the new energy vehicle air conditioning component test run detection.

[0095] The new energy vehicle air conditioning component without defects is determined by subsequent use authentication that the new energy vehicle air conditioning component has no defects, and the audio data of the new energy vehicle air conditioning component test run detection without defects is obtained as follows:

[0096] All the audio data of the new energy vehicle air conditioning component test run detection is encoded and saved in the database;

[0097] The use of the subsequent new energy vehicle air conditioning component is monitored, and after reaching the set use time, it is determined that there is no defect, the audio data of the new energy vehicle air conditioning component test run detection with defects is deleted, and the remaining is the audio data of the new energy vehicle air conditioning component test run detection without defects.

[0098] After obtaining the audio data of the new energy vehicle air conditioning component detection, the audio data needs to be processed and analyzed, and according to the analysis result, it can be determined whether the new energy vehicle air conditioning component has defects, and the process is based on the defect analysis module.

[0099] The defect analysis module is used to analyze the audio data of the historical state new energy vehicle air conditioning component test run detection, set standard audio features and standard intervals, then process the real-time collected audio data of the new energy vehicle air conditioning component test run detection, compare the processed audio features with the standard audio features, calculate the real-time difference ratio, and compare the real-time difference ratio with the standard interval to judge whether the new energy vehicle air conditioning component has defects;

[0100] The analysis of the audio data of the historical state new energy vehicle air conditioning component test run detection includes analyzing the main features and auxiliary spectrum features, and the steps of analyzing the main features are as follows:

[0101] The obtained historical state new energy automobile air conditioning component commissioning detection audio data is divided into N frames, and the time length of each frame is 20-40 milliseconds;

[0102] The divided N frames of audio data are converted into frequency domain representation by Fourier transformation;

[0103] The Fourier transformation into frequency domain representation is a short-time Fourier transformation, and the process includes framing, windowing and Fourier transformation;

[0104] The converted signal spectrum is filtered using a Mel filter;

[0105] The signal output by the Mel filter is subjected to logarithmic operation to convert it into a Mel frequency coefficient;

[0106] The Mel frequency coefficient is subjected to discrete cosine transformation to obtain a preliminary feature coefficient;

[0107] The preliminary feature coefficient is subjected to energy item removal and difference item processing to obtain an MFCC feature vector;

[0108] The above steps are the feature extraction commonly used in existing speech recognition technology, and the present scheme does not improve them, so the specific extraction and transformation will not be described in detail.

[0109] The auxiliary spectral features include any A of spectral energy, spectral centroid, spectral entropy, spectral peak, spectral attenuation, spectral flux, spectral kurtosis, spectral slope, spectral skewness, spectral spread and spectral roll-off point, and 3≤A≤5.

[0110] Other sound features can also be used.

[0111] The higher the value of A, the more accurate the analysis result, but the amount of data for analysis will greatly increase. Based on the current requirements, the value range of A is 3-5, which is the best choice. When the value range of A is 3-5, the analysis time can be guaranteed to be between 10-20 seconds.

[0112] The steps of setting standard audio data are as follows:

[0113] The main features and auxiliary spectral feature data of the K groups of historical state new energy automobile air conditioning component commissioning detection audio data calculated recently are obtained;

[0114] K is 100-200. The K groups of data are close to or the same as the current production process, so they are more accurate for comparison. If historical data that is too long ago is used, the difference is likely to be large, and the comparison effect is not good.

[0115] The average data of the K groups of main features and auxiliary spectral feature data is calculated to obtain the average of the N MFCC feature vectors and the average of the A groups of auxiliary spectral features, and the average data is summarized to obtain the standard audio features, i.e., the standard audio features contain N+A data.

[0116] The difference ratios of the K groups of data and the average data are respectively calculated to obtain the maximum and minimum of the single feature difference ratio and the maximum and minimum of the comprehensive feature difference ratio, thereby obtaining the single standard interval and the comprehensive standard interval.

[0117] For example, the average calculated according to the array of a feature is 55, the maximum data of which is 64, and the minimum data of which is 47, the difference ratios between 64 and 55 and between 47 and 55 are respectively calculated, which are 16.3 and -14.5, and then the absolute values of the calculation results are compared, 16.3 is greater than 14.5, and thus the single standard interval is (0, 16.3).

[0118] The formula for calculating the single feature difference ratio is as follows:

[0119]

[0120] In the formula, Diffd is the single feature difference ratio, Tz is the audio feature data, is the average of the K groups of audio feature data;

[0121] When the equipment is abnormal, the position will be abnormal, which will affect the normal operation, and the most obvious case is vibration, and the vibration will produce sound, so that the characteristics of the sound change, thus Tz is obviously larger or smaller, so that the calculated Diffd is larger. At present, the production technology is relatively mature, therefore, the case of too large single feature difference ratio is rare, and it is basically an individual case, which normally occurs during the adaptation period of equipment replacement or process replacement.

[0122] The formula for calculating the comprehensive feature difference ratio is as follows:

[0123]

[0124] In the formula, Diffz is the comprehensive feature difference, Tzz i is the i-th MFCC feature vector data, is the average calculated based on all i-th MFCC feature vector data of the K groups of historical data, Xsz is the weight coefficient of the MFCC feature vector data difference ratio, Tzf x is the x-th auxiliary spectral feature data, is the average of all x-th auxiliary spectral feature data of the K groups of historical data, Xsf xThe weight coefficient of the xth auxiliary spectrum feature data difference value ratio.

[0125] This step is mainly due to the different proportions of different features, in order to better reflect the comprehensive situation, the need for weighting processing, weighting judgment.

[0126] After a long time of use, it is found that most of the defects are found by comprehensive feature difference ratio, mainly due to mature technology, generally no big defects in a single position, therefore, generally no obvious change in features, but through multiple feature analysis, amplification, it can also judge some small defects, therefore, in actual use, the calculation of the comprehensive feature difference ratio is required.

[0127] The real-time difference ratio includes real-time single difference ratio and real-time comprehensive difference ratio, and the comparison between the real-time difference ratio and the standard interval includes the comparison between the real-time single difference ratio and the single standard interval, and the comparison between the real-time comprehensive difference ratio and the comprehensive standard interval.

[0128] When judging whether the new energy automobile air conditioning component has defects or not,

[0129] If the real-time single difference value is located in the single standard interval, and the real-time comprehensive difference ratio is located in the comprehensive standard interval, it is determined that there is no defect.

[0130] Other conditions are all defects.

[0131] The real-time single difference value is located outside the single standard interval, which is usually a large local structure defect, and the real-time comprehensive difference ratio is located outside the comprehensive standard interval, which is usually a plurality of small defects.

[0132] When the defect is analyzed, the cause and degree of the defect need to be further analyzed, so that the relevant personnel can summarize the cause of the defect, so that the subsequent further adjustment scheme can be facilitated, and the equipment can be long-term and stable. Run, and further analyze the defect based on the abnormal analysis module.

[0133] The present application provides a new energy automobile air conditioning component production management system and management method, which can collect new energy automobile air conditioning component historical detection data, and can further analyze the historical data, research the standard audio feature and standard interval of the standard new energy automobile air conditioning component, so as to facilitate the comparison between the processed real-time collected new energy automobile air conditioning component historical detection data and the standard audio feature and standard interval, and quickly judge whether there is a defect or not, and facilitate the relevant technical personnel to understand the situation of the produced new energy automobile air conditioning component, which is convenient to use, has good overall use effect, and has good use prospect.

[0134] Anomaly analysis module: used to start when receiving a defective signal, calculate the real-time defect anomaly level, and predict the defect cause according to the audio performance, and mark the new energy vehicle air conditioner part assembly welding position with defects according to the defect cause, and summarize to form a predicted defect list;

[0135] The formula for calculating the real-time defect anomaly level is as follows:

[0136]

[0137] In the formula, Def is the calculated real-time defect anomaly level, Def=ROUNDUP() is the upward rounding function, Defd is the distance between the real-time single difference value proportion and the single standard interval, Blxsd is the distance between the real-time single difference value proportion and the single standard interval, Defz is the distance between the real-time comprehensive difference value proportion and the comprehensive standard interval, Blxsz is the distance between the real-time comprehensive difference value proportion and the comprehensive standard interval, B is the distance difference corresponding to the preset corresponding anomaly level, L is the number of defects in the N MFCC feature vectors, cz i is the real-time single difference value proportion of the i-th defect MFCC feature vector, Maxz i is the maximum difference value proportion of the i-th defect MFCC feature vector, Minz i is the minimum difference value proportion of the i-th defect MFCC feature vector, Defd1 is the distance between the auxiliary spectral feature data and the corresponding single standard interval, Z is the number of defects in the auxiliary spectral feature data, czf j is the real-time single difference value proportion of the j-th defect auxiliary spectral feature data, Maxf j is the maximum difference value proportion of the j-th defect auxiliary spectral feature data, Minf j is the minimum difference value proportion of the j-th defect auxiliary spectral feature data, Xsf j is the weight coefficient of the j-th defect auxiliary spectral feature data difference value proportion, Zcz is the real-time comprehensive difference value proportion, Zmax is the maximum value of the comprehensive standard interval, and Zmin is the minimum value of the comprehensive standard interval.

[0138] The above is the anomaly level data based on audio data analysis.

[0139] When analyzing abnormal data, there are many possibilities for the causes of the abnormal performance. At this time, all the causes are listed, and which causes correspond to the assembly welding position (the parts are detected during production, so the assembled parts are all defect-free parts, therefore, the defects are caused during assembly or welding) are marked, and a predicted defect list is summarized.

[0140] For example, the air conditioner heat exchanger of the automobile, analyze the cause of the leakage noise, it may not be fully aligned during assembly, it may be the heat exchange zone welding effect is not good, there is a gap in the welding position, at this time will be marked out the assembly and welding position.

[0141] The sound collection source can also be increased at the same time. The sound collected from different sources is different, and the data reflected is also different. The approximate position of the abnormality can be judged (for the case where the real-time single difference value is located outside the single standard range), so that subsequent detection is facilitated. However, in this case, the amount of data analyzed is large, and more sound collection equipment is required, so the cost is also relatively high.

[0142] According to the audio performance, the defect reason is predicted, and the new energy automobile air conditioner part assembly welding position with defects is marked according to the defect reason, and the predicted defect list is formed.

[0143] When predicting the defect reason according to the audio performance, all defect characteristic data is input into the database, 2-3 groups of historical abnormal data closest to each defect characteristic data are found in the database, and the defect reason recorded in the defect abnormal data is called.

[0144] For example, in the case of real-time single difference value located outside the single standard range, there are multiple characteristics that do not meet the requirements. At this time, the abnormal data of each characteristic needs to be searched, for example, there are 2 abnormal characteristics, at this time, search twice, obtain 4-6 groups of historical abnormal data, and extract the defect reasons corresponding to the 4-6 groups of historical abnormal data.

[0145] When it belongs to the case where the real-time comprehensive difference value is located outside the comprehensive standard range, then directly search once, obtain 2-3 groups of historical abnormal data, and extract the defect reasons corresponding to the 2-3 groups of historical abnormal data.

[0146] When the possible defects and defect reasons are known, further analysis is needed to determine which defect is the specific cause. When the specific defect is clear, further analysis is needed to determine whether there is only this defect, so as to achieve comprehensive detection. The process is based on the comprehensive analysis module.

[0147] Comprehensive analysis module: used for obtaining picture data of new energy automobile air conditioner part assembly welding position corresponding to the predicted defect list, and analyzing the obtained picture data to determine whether there is a defect;

[0148] If there is a defect, the defect data is input into the database, and the defect anomaly level is predicted based on the defect data, the defect level difference between the predicted defect anomaly level and the real-time defect anomaly level is calculated, and the defect level difference is compared with the preset defect variation interval. If it is located inside the defect variation interval, the defect formation reason is found based on the defect data, and the defect is repaired;

[0149] If there is no defect or the defect level difference is located outside the defect variation interval, the new energy automobile air conditioner part assembly welding position on the predicted defect list is comprehensively detected, the defect area and the defect formation reason are obtained according to the comprehensive detection data, and the defect is repaired;

[0150] In this case, the defect is not obvious, and it is difficult to query the defect by using a simple picture analysis method. At this time, comprehensive detection is needed to analyze the defect. Comprehensive detection includes air tightness detection, ultrasonic flaw detection, X camera detection, magnetic powder detection, laser detection, infrared detection and other schemes. According to the defect reason of the part, the corresponding scheme can be selected.

[0151] The case that the defect level difference is located outside the defect variation interval is not only that there is a defect in this part, but also there are defects in other places, so further analysis is needed to analyze all the defects.

[0152] The scheme of predicting the defect anomaly level based on the defect data mainly plays the effect of checking whether it is missing, which can effectively avoid the incomplete defect detection caused by human negligence, has good detection effect, and has good use prospect.

[0153] The steps of analyzing the obtained picture data to judge whether there is a defect are as follows:

[0154] The obtained picture data is called and processed by using threshold segmentation method to remove the picture background;

[0155] The position where the defect may exist is enlarged and photographed by using fixed position, so the photographed picture is a standard picture. At this time, the picture does not need to be further cropped, and the whole use is convenient,

[0156] The picture after removing the background is denoised by using Gaussian filter, and then grayscale processing is performed;

[0157] Gaussian filter can better preserve the edge details of the image, avoid the image becoming blurred, and greatly improve the accuracy of subsequent image recognition.

[0158] Sauvola threshold method is used to binarize the pixel points of the grayscale processed picture, and the defect image on the picture is segmented;

[0159] The Sauvola threshold method is one of the commonly used schemes for image segmentation, which can identify relatively shallow defects, has high accuracy, avoids the case that small defects are not monitored, and has good overall use effect.

[0160] The segmented image is a defect image, and subsequent analysis and processing of the image can realize analysis of the defect.

[0161] When there is a defect image, the number of pixel points in the segmented image and the gray value of the pixel points in the image are obtained.

[0162] The formula for predicting the defect abnormality level based on the defect data is as follows:

[0163]

[0164] In the formula, Def1 is the predicted defect abnormality level, D is the number of defect images, hdc i is the gray data of the i-th defect image, hsxs i is the conversion ratio, E is the total number of pixel points of the i-th defect image, hd j is the gray value of the j-th pixel point in the i-th defect image, hdb j is the standard value set for the i-th defect image, and bdz is the difference value, 0

[0165] For example, there are multiple predicted defect positions, so there are multiple defect images taken, and the images without defects are filtered out, leaving D images. At this time, each defect image needs to be analyzed to calculate the abnormality level of each defect image, and then summarized. For example, there are 3 defect images, and the defect levels calculated for each defect image are 2.1, 2.4, and 1.8. At this time, the conversion processing is performed (the importance of each defect position is different, and the reaction caused by the defect is different, for example, a defect exists in a curved position of a pipeline, and the sound caused by the defect is greater than that in a straight position, so the corresponding conversion needs to be performed).

[0166] The defect level difference Defc = Def-Def1, and the preset defect variation range is where Def max is the set maximum defect abnormality level, and β is a constant data, 25≤β≤35.

[0167] The preset defect variation range is the calculation error, which is within a certain range, but when the defect is outside the variation range, it is obviously an abnormal situation and needs to be analyzed again.

[0168] The application provides a production management system and management method for new energy automobile air conditioner components, which can quickly analyze and judge the assembled new energy automobile air conditioner components, judge whether defects exist, further analyze and predict defect causes, automatically call pictures corresponding to the defect causes, further analyze the pictures, and judge the defect forming causes. The sequential judgment mode can reduce the analysis amount of pictures, effectively reduce the data processing amount, and is comprehensive detection, can reflect the cooperation between parts and the welding effect, further reflect the assembly capacity of enterprises for parts, has good use effect, and has good use prospect.

[0169] When defects are detected, in order to ensure the qualified rate, it is necessary to verify whether the defects belong to normal defects or occasional defects. When the defects belong to normal defects, artificial further adjustment is needed, and when the defects belong to occasional defects, no treatment is needed. At this time, most of them are fluctuations occasionally caused by long-time work of equipment or misoperation of related personnel. The occurrence frequency is relatively low, so no treatment is needed. The above process is based on a normal processing module.

[0170] The normal processing module is used for randomly extracting defect areas of W new energy automobile air conditioner parts in the batch for detection, judging whether defects exist, and calculating a defect rate.

[0171] If the defect rate is greater than a set value, the corresponding processing scheme is adjusted according to the defect cause.

[0172] If the defect rate is less than the set value, no treatment is needed.

[0173] When the defect area is detected, the detection range is the center of the defect area, the detection area is γ times the defect area, and γ>3. Normally, in order to reduce the data amount, γ is usually <5.

[0174] The calculated defect rate S q W is the number of new energy automobile air conditioner parts in which the position has defects. The set value is δ, and 3%<δ<5%. The lower δ is, the higher the requirement is.

[0175] For example, 20 parts are randomly inspected, and the number of parts with defects is 2. The calculated Der is 10%, which is obviously greater than the set value, indicating that it belongs to normal defects. At this time, the corresponding adjustment scheme is needed to reduce the existence of the defects.

[0176] The application provides a new energy automobile air conditioner component production management system and management method, which analyzes and judges new energy automobile air conditioner components, can analyze the defect formation reason and position of the new energy automobile air conditioner components, and detects all new energy automobile air conditioner components of the same batch in a targeted manner, can help relevant technical personnel understand whether the new energy automobile air conditioner components belong to normal problems, so that relevant management personnel can formulate corresponding production strategies, so that the qualified rate of the produced new energy automobile air conditioner components is higher, the use effect is good, and the application scenarios are good.

[0177] The weight coefficient involved in the above formula is determined by the coefficient of variation method. The coefficient of variation method is a method of weighting each index according to the variation degree of the current value and the target value of each evaluation index. If the numerical difference of an index is large, the index can clearly distinguish each evaluated object, indicating that the index has rich distinguishing information, and thus the index should be given a larger weight. On the contrary, if the numerical difference of each evaluated object on an index is small, the index has weak ability to distinguish each evaluated object, and thus the index should be given a smaller weight. This method directly uses the information contained in each index to obtain the weight of the index by calculation, and thus is objective.

[0178] Embodiment 2

[0179] A new energy automobile air conditioner component production management method comprises the following steps:

[0180] Collect historical state and real-time new energy automobile air conditioner component data, the real-time new energy automobile air conditioner component data comprising audio data of new energy automobile air conditioner component trial operation detection and picture data of new energy automobile air conditioner part assembly welding position after detection, and the historical state new energy automobile air conditioner component data being audio data of new energy automobile air conditioner component trial operation detection without defects;

[0181] The above step is to collect relevant data, which lays the foundation for subsequent analysis and processing.

[0182] The audio data of the historical state new energy automobile air conditioner component trial operation detection is analyzed, the standard audio feature and the standard interval are set, then the audio data of the real-time new energy automobile air conditioner component trial operation detection is processed, the processed audio feature is compared with the standard audio feature, the real-time difference ratio is calculated, and the real-time difference ratio is compared with the standard interval to judge whether the new energy automobile air conditioner component has defects;

[0183] The above step is to analyze and judge the data to determine whether there are defects.

[0184] Start when receiving the defect signal, calculate the real-time defect abnormality level, and predict the defect cause according to the audio performance, and mark the new energy vehicle air conditioner part assembly welding position with defects according to the defect cause, and summarize to form a predicted defect list;

[0185] Obtain the picture data of the new energy vehicle air conditioner part assembly welding position corresponding to the predicted defect list, and analyze the obtained picture data to determine whether there is a defect;

[0186] If there is a defect, input the defect data into the database, and predict the defect abnormality level based on the defect data, calculate the defect level difference between the predicted defect abnormality level and the real-time defect abnormality level, and compare the defect level difference with the preset defect variation interval. If it is located inside the defect variation interval, find the defect formation reason based on the defect data, and repair the defect;

[0187] If there is no defect or the defect level difference is located outside the defect variation interval, the new energy vehicle air conditioner part assembly welding position on the predicted defect list is comprehensively detected, the defect area and the defect formation reason are obtained according to the comprehensive detection data, and the defect is repaired;

[0188] The above steps are to determine whether the defect is completely detected.

[0189] Randomly extract the defect area of the batch of W new energy vehicle air conditioner parts for assembly for detection, determine whether there is a defect, and calculate the defect rate;

[0190] If the defect rate is greater than the set value, adjust the corresponding processing scheme according to the defect cause;

[0191] If the defect rate is less than the set value, no processing is performed.

[0192] The above steps are to determine whether the defect is a normal defect.

[0193] In the present application, the several formulas involved are calculated by taking the numerical value after dimensionless, and the establishment of the formula is obtained by software simulation of a formula of the nearest real situation by collecting a large amount of data. Part of the coefficients or weights in the formula are set by the person skilled in the art according to the actual situation, so this will not be described here.

[0194] The above-described embodiments can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented by software, the above-described embodiments can be implemented in whole or in part in the form of a computer program product. A person of ordinary skill in the art can be aware that units and algorithm steps of the examples described in connection with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether the functions are performed by hardware or software depends on the specific application and design constraints of the technical solutions.

[0195] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, and can be located in one place or distributed on multiple network units. Part or all of the units can be selected to achieve the purpose of the embodiments.

[0196] The above describes only specific embodiments of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical scope disclosed by the present application, which should be covered within the protection scope of the present application.

Claims

1. A production management system for new energy automobile air conditioning components, characterized in that: The method comprises the following steps: A data acquisition module is used to collect historical state and real-time new energy vehicle air conditioner component data, wherein the real-time new energy vehicle air conditioner component data comprises audio data of new energy vehicle air conditioner component trial operation detection and picture data of new energy vehicle air conditioner part assembly welding positions after detection, and the historical state new energy vehicle air conditioner component data is audio data of new energy vehicle air conditioner component trial operation detection without defects; A defect analysis module is used to analyze the audio data of new energy vehicle air conditioner component trial operation detection in the historical state, set standard audio features and a standard interval, then process the audio data of new energy vehicle air conditioner component trial operation detection collected in real time, compare the processed audio features with the standard audio features, calculate a real-time difference ratio, and compare the real-time difference ratio with the standard interval to determine whether the new energy vehicle air conditioner component has defects; The steps of setting the standard audio data are as follows: Obtain the main features and auxiliary spectral feature data of the audio data of the K groups of new energy vehicle air conditioner component trial operation detection calculated in the historical state; Calculate the average data of the K groups of main features and auxiliary spectral feature data to obtain the average of N MFCC feature vectors and the average of A groups of auxiliary spectral features, and aggregate the average data to obtain the standard audio features; Calculate the difference ratio of each group of data with the average data, obtain the maximum and minimum of the single feature difference ratio, and obtain the maximum and minimum of the comprehensive feature difference ratio to obtain the single standard interval and the comprehensive standard interval; The formula for calculating the single feature difference ratio is as follows: wherein, is a single feature difference ratio, is audio feature data, is an average value of K groups of audio feature data; The formula for calculating the comprehensive feature difference ratio is as follows: wherein, is the comprehensive feature difference value, is the ith MFCC feature vector data, is the average of all ith MFCC feature vector data based on K sets of historical data, is the weight coefficient of the MFCC feature vector data difference value ratio, is the xth auxiliary spectral feature data, is the average of all xth auxiliary spectral feature data based on K sets of historical data, is the weight coefficient of the xth auxiliary spectral feature data difference value ratio; An anomaly analysis module is used to start when a defect signal is received, calculate a real-time defect anomaly level, predict the defect cause according to the audio performance, mark the new energy vehicle air conditioner part assembly welding position with defects according to the defect cause, and aggregate to form a predicted defect list; The formula for calculating the real-time defect anomaly level is as follows: wherein, is a real-time defect abnormality level calculated, is a rounding-up function, is a distance between the real-time single-item difference ratio and the single-item standard interval, is a proportional coefficient of the distance between the real-time single-item difference ratio and the single-item standard interval, is a distance between the real-time comprehensive difference ratio and the comprehensive standard interval, is a proportional coefficient of the distance between the real-time comprehensive difference ratio and the comprehensive standard interval, B is a preset distance difference value corresponding to the respective abnormality level, and L is a number of MFCC feature vectors with defects, is a real-time single-item difference ratio of the i-th defect MFCC feature vector, is a maximum difference ratio of the i-th defect MFCC feature vector corresponding to the single-item standard interval, is a minimum difference ratio of the i-th defect MFCC feature vector corresponding to the single-item standard interval, is a distance between the auxiliary spectral feature data and the corresponding single-item standard interval, and Z is a number of auxiliary spectral feature data with defects, is a real-time single-item difference ratio of the j-th defect auxiliary spectral feature data, is a maximum difference ratio of the j-th defect auxiliary spectral feature data corresponding to the single-item standard interval, is a minimum difference ratio of the j-th defect auxiliary spectral feature data corresponding to the single-item standard interval, is a weight coefficient of the difference ratio of the j-th defect auxiliary spectral feature data, is a real-time comprehensive difference ratio, is a maximum value of the comprehensive standard interval, is a minimum value of the comprehensive standard interval. A comprehensive analysis module is used to obtain picture data of the new energy vehicle air conditioner part assembly welding position corresponding to the predicted defect list, analyze the obtained picture data, and determine whether there are defects; If there are defects, input the defect data into a database, predict the defect anomaly level based on the defect data, calculate the defect level difference between the predicted defect anomaly level and the real-time defect anomaly level, compare the defect level difference with a preset defect variation interval, if it is within the defect variation interval, find the defect formation cause based on the defect data, and repair the defect; If there are no defects or the defect level difference is outside the defect variation interval, perform comprehensive detection on the new energy vehicle air conditioner part assembly welding positions on the predicted defect list, obtain the defect area and the defect formation cause according to the comprehensive detection data, and repair the defect; A normal processing module is used to randomly select W new energy vehicle air conditioner part assembly defect areas in the batch for detection, determine whether there are defects, and calculate a defect rate; If the defect rate is greater than a set value, adjust the corresponding processing scheme according to the defect cause. If the defect rate is less than the set value, no processing is performed.

2. The production management system of a new energy automobile air conditioning component according to claim 1, characterized in that: The audio data obtained from the historical state new energy automobile air conditioning component test run detection is analyzed, and the main features and auxiliary spectral features are analyzed. The steps of analyzing the main features are as follows: The audio data obtained from the historical state new energy automobile air conditioning component test run detection is divided into N frames; The N frames of audio data after division are converted into frequency domain representation through Fourier transformation; The signal spectrum after conversion is filtered using a Mel filter; The signal output by the Mel filter is subjected to logarithmic operation to convert it into Mel frequency coefficients; The Mel frequency coefficients are subjected to discrete cosine transformation to obtain preliminary feature coefficients; The preliminary feature coefficients are subjected to energy item removal and difference item addition processing to obtain MFCC feature vectors; The auxiliary spectral features include any A of spectral energy, spectral centroid, spectral entropy, spectral peak, spectral attenuation, spectral flux, spectral kurtosis, spectral slope, spectral skewness, spectral spread, and spectral roll-off point.

3. The production management system of a new energy automobile air conditioning component according to claim 2, characterized in that: The real-time difference ratio includes real-time single-item difference ratio and real-time comprehensive difference ratio. Comparing the real-time difference ratio with the standard interval includes comparing the real-time single-item difference ratio with the single-item standard interval and comparing the real-time comprehensive difference ratio with the comprehensive standard interval. When judging whether the new energy automobile air conditioning component has defects; If the real-time single-item difference is within the single-item standard interval and the real-time comprehensive difference ratio is within the comprehensive standard interval, it is determined that there are no defects; Other cases are all defects.

4. The production management system of a new energy automobile air conditioning component according to claim 3, characterized in that: When predicting the defect cause based on audio performance, all defect feature data is input into the database. In the database, 2-3 groups of historical abnormal data closest to each defect feature data are found, and the defect causes recorded in the defect abnormal data are retrieved.

5. The production management system of a new energy automobile air conditioning component according to claim 4, characterized in that: The steps of analyzing the obtained picture data to determine whether there are defects are as follows: The obtained picture data is retrieved, and threshold segmentation method is used to process the picture data to remove the picture background; Gaussian filtering is used to denoise the picture after removing the background, and then grayscale processing is performed; Sauvola threshold method is used to binarize the pixel points of the grayscale processed picture, and the defect image on the picture is segmented; The segmented image is the defect image; When there is a defect image, the number of pixel points in the segmented image and the grayscale value of the pixel points in the image are obtained. 6.The production management system of a new energy automobile air conditioning component according to claim 5, characterized in that: The formula for predicting the defect abnormal level based on the defect data is as follows: wherein, D is the number of defect images, is the gray data of the i-th defect image, is the conversion ratio, is the total number of pixel points of the i-th defect image, is the gray value of the j-th pixel point in the i-th defect image, is the standard value set for the i-th defect image, is the difference fluctuation value, ; Defect level difference Defc = Def - Defl, and a preset defect variation range is wherein is a set maximum defect abnormality level, is a constant data, . 7.The production management system of a new energy automobile air conditioning component according to claim 6, characterized in that: When detecting the defect area, the detection range is centered on the defect area, and the detection area is γ times the defect area, γ>3; Calculated defect rate , W is the number of new energy automobile air conditioning parts with defects in this position, and the set value is δ, 3% < δ < 5%.

8. A production management method of a new energy automobile air conditioning component, using the system of any one of claims 1 to 7, characterized in that: The steps are as follows: Collect historical state and real-time new energy automobile air conditioning component data, real-time new energy automobile air conditioning component data includes audio data of new energy automobile air conditioning component test run detection and picture data of new energy automobile air conditioning part assembly welding position after detection, historical state new energy automobile air conditioning component data is audio data of new energy automobile air conditioning component test run detection without defects; The audio data of the historical state new energy automobile air conditioning component test run detection is analyzed, the standard audio features and the standard interval are set, then the audio data of the real-time collected new energy automobile air conditioning component test run detection is processed, and the processed audio features are compared with the standard audio features, the real-time difference ratio is calculated, and the real-time difference ratio is compared with the standard interval, whether the new energy automobile air conditioning component exists defects is judged; When receiving the defect signal, the real-time defect abnormality level is calculated, the defect reason is predicted according to the audio performance, the new energy automobile air conditioning part assembly welding position with defects is marked according to the defect reason, and the predicted defect list is formed by summarizing; The picture data of the new energy automobile air conditioning part assembly welding position corresponding to the predicted defect list is obtained, and the obtained picture data is analyzed to determine whether there is a defect; If there is a defect, the defect data is input into the database, the defect abnormality level is predicted based on the defect data, the defect level difference between the predicted defect abnormality level and the real-time defect abnormality level is calculated, and the defect level difference is compared with the preset defect change interval; if it is located in the defect change interval, the defect forming reason is found based on the defect data, and the defect is repaired; If there is no defect or the defect level difference is located outside the defect change interval, the new energy automobile air conditioning part assembly welding position on the predicted defect list is comprehensively detected, the defect area and the defect forming reason are obtained according to the comprehensive detection data, and the defect is repaired; Randomly extract the defect area of the batch W new energy automobile air conditioning part assembly for detection, judge whether there is a defect, and calculate the defect rate; If the defect rate is greater than the set value, adjust the corresponding processing scheme according to the defect reason; If the defect rate is less than the set value, no processing is performed.

Images