A method for constructing an intelligent product risk prediction model based on machine learning

Abstract

The present invention discloses a method for constructing an intelligent product risk prediction model based on machine learning, and relates to the technical field of risk prediction management. The method for constructing an intelligent product risk prediction model based on machine learning includes the following steps: production noise interference assessment; sensor processing capability assessment; risk assessment; risk prediction classification. The present invention obtains a sensor processing capability assessment value by combining sensor processing capability data and automobile production noise interference value and determines whether to adjust the sensor capability, then obtains an automobile risk assessment value by automobile production abnormal state data and sensor processing capability assessment value, and finally performs risk prediction classification based on the constructed automobile risk prediction model, thereby achieving the effect of improving the data accuracy of automobile manufacturing risk prediction based on machine learning, and solving the problem of inaccurate data in automobile manufacturing risk prediction based on machine learning in the prior art.

Description

Technical Field

[0001] The present invention relates to the technical field of risk prediction management, and in particular to a method for constructing an intelligent product risk prediction model based on machine learning. Background Art

[0002] Machine learning, as an artificial intelligence technology, enables machines to make decisions and predictions autonomously by training algorithms and analyzing data. It has the ability to extract patterns and regularities from very large data sets, which can be used to predict future trends and outcomes. In recent years, machine learning technology has made significant progress and has been widely used in many fields. In terms of risk prediction, machine learning technology has the following significant advantages: machine learning algorithms can extract useful features from large amounts of data and build high-precision prediction models. Compared with traditional methods, machine learning models can usually provide more accurate prediction results; machine learning technology can automatically learn regularities and patterns from data, reducing the need for manual intervention and expert judgment. This makes the risk assessment process more automated and intelligent, and improves the efficiency and accuracy of assessment.

[0003] The existing risk prediction model construction method mainly analyzes historical data and related indicators and uses statistical methods to infer possible future risks. Statistical analysis can use various statistical models, such as regression analysis and time series analysis, to build risk prediction models. Automobile manufacturers use risk prediction models to monitor and analyze quality data in the production process, identify potential quality problems in a timely manner, and take measures to improve them.

[0004] For example, the training method, prediction method, device, medium and equipment of the risk prediction model announced in the invention patent with announcement number: CN110674979B include: obtaining training samples with time series, and inputting the training samples into the risk prediction model; obtaining derivative features with time series output by the feature derivation layer of the risk prediction model, wherein the derived features are obtained by deriving features of the training samples by the feature derivation layer; calculating the time series loss function value of the feature derivation layer according to the distribution difference of the derived features at different time points in the time series; determining the objective function value of the risk prediction model according to the time series loss function value, so as to train the risk prediction model according to the objective function value.

[0005] For example, the risk prediction method, device, computer equipment and storage medium based on machine learning announced by the invention patent with announcement number: CN109543925B include: obtaining first risk data of training samples; preprocessing the first risk data to obtain multiple basic risk factors; identifying related samples of the training samples, crawling second risk data of the related samples; preprocessing the second risk data to obtain multiple related risk factors; screening the basic risk factors and related risk factors to obtain multiple effective risk factors; training the machine learning model based on the effective risk factors to obtain a risk prediction model; monitoring whether the monitored object has risky behavior based on the risk prediction model.

[0006] However, in the process of implementing the technical solution of the invention in the embodiments of the present application, the present application found that the above technology has at least the following technical problems:

[0007] In the prior art, since environmental interference in smart car manufacturing may enter the data transmission system through transmission lines or sensors, causing data distortion or bit errors, it leads to inaccurate data for automobile manufacturing risk prediction based on machine learning. Summary of the invention

[0008] The embodiments of the present application provide a method for constructing an intelligent product risk prediction model based on machine learning, thereby solving the problem of inaccurate data in automobile manufacturing risk prediction based on machine learning in the prior art, and improving the accuracy of data in automobile manufacturing risk prediction based on machine learning.

[0009] The embodiment of the present application provides a method for constructing a smart product risk prediction model based on machine learning, comprising the following steps: S1, obtaining an automobile production noise interference value based on the automobile production noise data of the specified automobile production line, wherein the automobile production noise interference value is used to evaluate the influence of the noise of the specified automobile production line on the transmission of automobile production data to a preset data monitoring center; S2, judging whether to perform interference adjustment based on the automobile production noise interference value, obtaining the automobile production noise interference value after the interference adjustment and combining it with the obtained sensor processing capability data to obtain a sensor processing capability evaluation value, wherein the sensor processing capability evaluation value is used to evaluate the processing capability of the preset data monitoring center for the transmitted automobile production noise data in the intelligent automobile manufacturing process; S3, judging whether to perform sensor capability adjustment based on the sensor processing capability evaluation value, obtaining an automobile risk evaluation value based on the obtained automobile production abnormal state data and combining it with the sensor processing capability evaluation value after the sensor capability adjustment, wherein the automobile risk evaluation value is used to evaluate the automobile production abnormality in the intelligent automobile manufacturing process; S4, performing risk prediction classification based on the constructed automobile risk prediction model, wherein the risk prediction classification is used to reflect the risk level of the specified automobile production line, and the automobile risk prediction model represents a model constructed by a machine learning algorithm for outputting risk prediction classification results.

[0010] Furthermore, the automobile production noise data includes airflow noise signal power, electromagnetic interference noise signal power and signal average power; the sensor processing capability data includes actual throughput and actual response time; the actual throughput indicates the amount of data corresponding to the automobile production data processed and transmitted by the preset data monitoring center per unit time; the actual response time indicates the time for the automobile production data to be transmitted to the preset data monitoring center; the automobile production abnormal status data includes the total automobile repair time and the total number of automobile failures.

[0011] Furthermore, the automobile production noise data is obtained by measuring the noise conditions of a designated automobile production line within a preset time period through a noise acquisition device; the noise acquisition device includes a sound level meter, an electromagnetic interference tester and a power meter; the sensor processing capacity data is obtained by measuring the sensor processing conditions of a designated automobile production line within a preset time period through a capacity processing device; the capacity processing device includes a network data flow analyzer and a response time tester; the automobile production abnormal status data is obtained by measuring the automobile abnormal conditions of a designated automobile production line within a preset time period through an abnormality acquisition device; the abnormality acquisition device includes a timer and a counter.

[0012] Furthermore, the specific process of obtaining the automobile production noise interference value based on the automobile production noise data of the designated automobile production line is as follows: the sum of the airflow noise signal power and the preset airflow noise power threshold obtained from the database is ratioed to twice the preset airflow noise power threshold to obtain the airflow noise power error degree value; the electromagnetic interference noise signal power and the preset electromagnetic interference noise power threshold obtained from the database are ratioed to twice the preset electromagnetic interference noise power threshold to obtain the electromagnetic interference noise power error degree value; the signal average power and the airflow noise signal power are ratioed to obtain the airflow signal-to-noise ratio; the telecommunication signal-to-noise ratio is ratioed to obtain the electromagnetic interference noise signal-to-noise ratio. The electromagnetic signal-to-noise ratio is obtained by performing a ratio operation on the magnetic interference noise signal power of the signal average power; the airflow signal-to-noise ratio error degree value is obtained by performing a ratio operation on twice the preset airflow signal-to-noise ratio threshold obtained from the database and the sum of the airflow signal-to-noise ratio and the preset airflow signal-to-noise ratio threshold; the electromagnetic interference signal-to-noise ratio error degree value is obtained by performing a ratio operation on twice the preset electromagnetic interference signal-to-noise ratio threshold obtained from the database and the sum of the electromagnetic signal-to-noise ratio and the preset electromagnetic interference signal-to-noise ratio threshold; the automobile production noise interference value is obtained by combining the airflow noise power error degree value, the electromagnetic interference noise power error degree value, the airflow signal-to-noise ratio error degree value and the electromagnetic interference signal-to-noise ratio error degree value.

[0013] Furthermore, the limiting expression of the automobile production noise interference value is as follows:

[0014] ;

[0015] ;

[0016] ;

[0017] ;

[0018] ;

[0019] In the formula, It represents the automobile production noise interference value of the i-th automobile in the specified automobile production line in the m-th preset time period, , i represents the number of the car on the specified car production line, s represents the total number of cars on the specified car production line, , m represents the number of the preset time period, z represents the total number of preset time periods, It represents the airflow noise power error value of the i-th car in the specified car production line in the m-th preset time period, It represents the electromagnetic interference noise power error value of the i-th car in the specified car production line in the m-th preset time period, It represents the error degree of airflow signal-to-noise ratio of the i-th car in the specified car production line in the m-th preset time period. It represents the electromagnetic interference signal-to-noise ratio error value of the i-th car in the specified car production line in the m-th preset time period, represents the airflow noise signal power of the i-th car in the specified car production line in the m-th preset time period, represents the electromagnetic interference noise signal power of the i-th car in the specified car production line in the m-th preset time period, It represents the average signal power of the i-th car in the specified car production line in the m-th preset time period, Indicates the preset airflow noise power threshold, Indicates the preset electromagnetic interference noise power threshold, Indicates the preset airflow signal-to-noise ratio threshold. Indicates the preset electromagnetic interference signal-to-noise ratio threshold.

[0020] Furthermore, the specific process of obtaining the automobile production noise interference value after interference adjustment and combining it with the obtained sensor processing capability data to obtain the sensor processing capability evaluation value is as follows: the sum of the actual throughput and the preset throughput threshold obtained from the database is ratioed with twice the preset throughput threshold to obtain a throughput error degree value, and the throughput error degree value is used to reflect the deviation of the processing efficiency of the preset data monitoring center for the transmitted automobile production noise data; the sum of the actual response time and the preset response time threshold obtained from the database is ratioed with twice the preset response time threshold to obtain a response time error degree value, and the response time error degree value reflects the deviation of the processing speed of the preset data monitoring center for the transmitted automobile production noise data; the sensor processing capability evaluation value is obtained by combining the throughput error degree value, the response time error degree value and the automobile production noise interference value after interference adjustment.

[0021] Furthermore, the specific process of obtaining the automobile risk assessment value based on the acquired automobile production abnormal status data and combined with the sensor processing capability evaluation value after the sensor capability adjustment is as follows: the total automobile repair time and the preset automobile total repair time threshold obtained from the database are ratioed to obtain a repair time error degree value, and the repair time error degree value is used to reflect the deviation of the fault repair time of the specified automobile production line; the total number of automobile failures and the preset total number of automobile failures threshold obtained from the database are ratioed to obtain a failure number error degree value, and the failure number error degree value is used to reflect the deviation of the fault repair number of the specified automobile production line; the automobile risk assessment value is obtained by combining the repair time error degree value, the failure number error degree value and the sensor processing capability evaluation value after the sensor capability adjustment.

[0022] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:

[0023] 1. The sensor processing capability evaluation value is obtained by combining the acquired sensor processing capability data and the automobile production noise interference value after interference adjustment, and it is determined whether to adjust the sensor capability. Then, the automobile risk evaluation value is obtained by combining the automobile production abnormal status data and the sensor processing capability evaluation value after sensor capability adjustment. Finally, risk prediction classification is performed based on the constructed automobile risk prediction model, thereby improving the accuracy of the constructed automobile risk prediction model, and further improving the data accuracy of automobile manufacturing risk prediction based on machine learning, which effectively solves the problem of inaccurate data in automobile manufacturing risk prediction based on machine learning in the prior art.

[0024] 2. The automobile production noise interference value is obtained by combining the airflow noise power error degree value, the electromagnetic interference noise power error degree value, the airflow signal-to-noise ratio error degree value and the electromagnetic interference signal-to-noise ratio error degree value, and then the sensor processing capability evaluation value is obtained by combining the throughput error degree value, the response time error degree value and the automobile production noise interference value after interference adjustment. Finally, the automobile risk evaluation value is obtained by combining the repair time error degree value, the failure number error degree value and the sensor processing capability evaluation value after sensor capability adjustment, thereby achieving an improvement in the accuracy of risk prediction results, and then achieving an improvement in the reliability of risk prediction results.

[0025] 3. By judging whether the automobile production noise interference value is not higher than the preset production noise threshold obtained from the database, and then judging whether the sensor processing capability assessment value is not lower than the preset processing capability threshold obtained from the database, and finally judging whether the automobile risk assessment value is not higher than the preset risk threshold obtained from the database, the accuracy of obtaining prediction-related data is improved, thereby achieving the improvement of the effectiveness of obtaining risk prediction results. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 A flowchart of a method for constructing an intelligent product risk prediction model based on machine learning provided in an embodiment of the present application;

[0027] Figure 2 An overall flow chart provided for the embodiments of the present application;

[0028] Figure 3 A statistical chart of changes in the sensor processing capability evaluation value provided in an embodiment of the present application. DETAILED DESCRIPTION

[0029] The embodiment of the present application solves the problem of inaccurate data in automobile manufacturing risk prediction based on machine learning in the prior art by providing a method for constructing an intelligent product risk prediction model based on machine learning. The method obtains the automobile production noise interference value and determines whether to perform interference adjustment, then obtains the automobile production noise interference value after interference adjustment and combines it with sensor processing capability data to obtain a sensor processing capability evaluation value and determines whether to adjust the sensor capability. Then, the automobile production abnormal status data and the sensor processing capability evaluation value after the sensor capability adjustment are combined to obtain the automobile risk evaluation value. Finally, risk prediction classification is performed based on the constructed automobile risk prediction model, thereby improving the data accuracy of automobile manufacturing risk prediction based on machine learning.

[0030] The technical solution in the embodiment of the present application is to solve the problem of inaccurate data in automobile manufacturing risk prediction based on machine learning. The overall idea is as follows:

[0031] The obtained sensor processing capability evaluation value is used to determine whether to adjust the sensor capability. Then, the automobile risk assessment value is obtained by combining the automobile production abnormal status data and the sensor processing capability evaluation value after the sensor capability adjustment. Finally, risk prediction classification is performed based on the constructed automobile risk prediction model, thereby achieving the effect of improving the data accuracy of automobile manufacturing risk prediction based on machine learning.

[0032] In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0033] like Figure 1As shown, it is a flow chart of a method for constructing a smart product risk prediction model based on machine learning provided by an embodiment of the present application, the method comprising the following steps: S1, production noise interference assessment: obtaining an automobile production noise interference value based on the automobile production noise data of a specified automobile production line, the automobile production noise data is used to reflect the noise impact of the specified automobile production line, and the automobile production noise interference value is used to assess the impact of the noise of the specified automobile production line on the transmission of automobile production data to a preset data monitoring center; S2, sensor processing capability assessment: judging whether to perform interference adjustment based on the automobile production noise interference value, obtaining the automobile production noise interference value after interference adjustment and combining it with the obtained sensor processing capability data to obtain a sensor processing capability assessment value, and the sensor processing capability assessment value is used to assess the preset data monitoring in the intelligent automobile manufacturing process The processing capability of the measurement center for the transmitted automobile production noise data, and the interference adjustment is used to reduce the noise interference of the designated automobile production line; S3, risk assessment: based on the sensor processing capability assessment value, determine whether to adjust the sensor capability, and obtain the automobile risk assessment value based on the acquired automobile production abnormal status data and the sensor processing capability assessment value after the sensor capability adjustment. The automobile risk assessment value is used to evaluate the automobile production abnormality in the intelligent automobile manufacturing process. The sensor capability adjustment is used to improve the processing capability of the preset data monitoring center for the transmitted automobile production noise data; S4, risk prediction classification: risk prediction classification is performed based on the constructed automobile risk prediction model. The risk prediction classification is used to reflect the risk level of the designated automobile production line. The automobile risk prediction model represents a model constructed by a machine learning algorithm for outputting risk prediction classification results.

[0034] like Figure 2 As shown, it is an overall flow chart provided by the embodiment of the present application. There is a close connection and mutual influence between the automobile production noise interference value, the sensor processing capability assessment value and the automobile risk assessment value. The higher the automobile production noise interference value, the more likely it is that the sensor data transmission error or loss will be caused, affecting the integrity and accuracy of the data. This will increase the difficulty of sensor processing capability assessment, thereby reducing the sensor processing capability assessment value. The higher the automobile risk assessment value means that there are more abnormal conditions in the production line, which may affect the quality and safety of the automobile.

[0035] Machine learning is the core of building an automobile risk prediction model. By training a large amount of historical data, the automobile risk assessment value that is not higher than the preset risk threshold and the corresponding automobile production noise interference value after interference adjustment, the sensor processing capability assessment value after sensor capability adjustment and the manually labeled expected risk classification results are input into the convolutional neural network model for training to obtain the automobile risk prediction model. The automobile risk assessment value that is no higher than the preset risk threshold and the corresponding automobile production noise interference value after interference adjustment and the sensor processing capability assessment value after sensor capability adjustment are input into the automobile risk prediction model to output the corresponding risk prediction classification results. The data accuracy of automobile manufacturing risk prediction based on machine learning is improved.

[0036] It should be added that the automobile production noise data includes the airflow noise signal power, the electromagnetic interference noise signal power and the signal average power; the sensor processing capability data includes the actual throughput and the actual response time; the actual throughput indicates the amount of data corresponding to the automobile production data processed and transmitted per unit time by the preset data monitoring center; the actual response time indicates the time it takes for the automobile production data to be transmitted to the preset data monitoring center; the automobile production abnormal status data includes the total automobile repair time and the total number of automobile failures.

[0037] Among them, the automobile production noise data is obtained by measuring the noise conditions of a designated automobile production line within a preset time period through a noise acquisition device; the noise acquisition device includes a sound level meter, an electromagnetic interference tester and a power meter; the airflow noise signal power is obtained through the sound level meter; the electromagnetic interference noise signal power is obtained through the electromagnetic interference tester; the signal average power is obtained through the power meter; the sensor processing capacity data is obtained by measuring the sensor processing conditions of a designated automobile production line within a preset time period through a capacity processing device; the capacity processing device includes a network data flow analyzer and a response time tester; the actual throughput is obtained through the network data flow analyzer; the actual response time is obtained through the response time tester; the automobile production abnormal status data is obtained by measuring the automobile abnormal conditions of a designated automobile production line within a preset time period through an abnormality acquisition device; the abnormality acquisition device includes a timer and a counter; the total repair time of the automobile is obtained through the timer; the total number of automobile failures is obtained through the counter.

[0038] It should be understood that at preset positions on the designated automobile production line, the airflow noise signal power within a preset time period is measured by a sound level meter deployed, and the airflow noise signal power of the cars on the designated automobile production line is obtained after the average value is calculated. The number of preset position points corresponds to the number of cars on the designated production line. The preset position points are manually marked by preset personnel according to the positions of the cars on the designated automobile production line. At preset positions on the designated automobile production line, the electromagnetic interference noise signal power within a preset time period is measured by an electromagnetic interference tester deployed, and the electromagnetic interference noise signal power of the cars on the designated automobile production line is obtained after the average value is calculated. The signal average power of the cars on the designated automobile production line is obtained after the signal power within a preset time period is measured by a power meter deployed at the preset position on the designated automobile production line. The method can improve the accuracy of data for automobile manufacturing risk prediction based on machine learning by measuring the response time within a preset time period through a response time tester deployed at a preset point on the designated automobile production line, thereby obtaining the response time. The method can also improve the accuracy of data for automobile manufacturing risk prediction based on machine learning by measuring the response time within a preset time period through a network data flow analyzer deployed at a preset point on the designated automobile production line, and then obtain the actual response time of the automobiles on the designated automobile production line after averaging the response time within a preset time period. The response time tester can record the transmission time of the data packet in the network to obtain the response time. The timer deployed on the designated automobile production line measures the total repair time of the automobile within the preset time period. The counter deployed on the designated automobile production line measures the total number of automobile failures within the preset time period.

[0039] Furthermore, the specific process of obtaining the automobile production noise interference value based on the automobile production noise data of the specified automobile production line is as follows: the sum of the airflow noise signal power and the preset airflow noise power threshold obtained from the database is ratioed to twice the preset airflow noise power threshold to obtain the airflow noise power error degree value (that is, the error degree value of the airflow noise power in the restriction expression of the automobile production noise interference value). ), the airflow noise power error degree value is used to reflect the deviation of the airflow noise interference of the specified automobile production line; the electromagnetic interference noise signal power and the preset electromagnetic interference noise power threshold obtained from the database are compared with 2 times the preset electromagnetic interference noise power threshold to obtain the electromagnetic interference noise power error degree value (that is, the limit expression of the automobile production noise interference value ), the electromagnetic interference noise power error degree value is used to reflect the deviation of the electromagnetic interference of the specified automobile production line; the airflow signal-to-noise ratio is obtained by ratio calculation of the signal average power and the airflow noise signal power, and the electromagnetic signal-to-noise ratio is obtained by ratio calculation of the magnetic interference noise signal power of the electrical signal average power; the airflow signal-to-noise ratio error degree value (i.e., the value in the limiting expression of the automobile production noise interference value) is obtained by ratio calculation of twice the preset airflow signal-to-noise ratio threshold obtained from the database and the sum of the airflow signal-to-noise ratio and the preset airflow signal-to-noise ratio threshold. ), the airflow signal-to-noise ratio error value is used to reflect the deviation of the airflow noise of the specified automobile production line on the signal quality; the electromagnetic interference signal-to-noise ratio error value (i.e., the value in the limiting expression of the automobile production noise interference value) is obtained by performing a ratio operation on the sum of the 2 times preset electromagnetic interference signal-to-noise ratio threshold obtained from the database and the electromagnetic signal-to-noise ratio and the preset electromagnetic interference signal-to-noise ratio threshold. ), the electromagnetic interference signal-to-noise ratio error degree value reflects the deviation of the electromagnetic interference noise of the specified automobile production line on the signal quality; the automobile production noise interference value is obtained by combining the airflow noise power error degree value, the electromagnetic interference noise power error degree value, the airflow signal-to-noise ratio error degree value and the electromagnetic interference signal-to-noise ratio error degree value.

[0040] Among them, the limiting expression of automobile production noise interference value is as follows:

[0041] ;

[0042] ;

[0043] ;

[0044] ;

[0045] ;

[0046] In the formula, It represents the automobile production noise interference value of the i-th automobile in the specified automobile production line in the m-th preset time period, , i represents the number of the car on the specified car production line, s represents the total number of cars on the specified car production line, , m represents the number of the preset time period, z represents the total number of preset time periods, It represents the airflow noise power error value of the i-th car in the specified car production line in the m-th preset time period, It represents the electromagnetic interference noise power error value of the i-th car in the specified car production line in the m-th preset time period, It represents the error degree of airflow signal-to-noise ratio of the i-th car in the specified car production line in the m-th preset time period. It represents the electromagnetic interference signal-to-noise ratio error value of the i-th car in the specified car production line in the m-th preset time period, represents the airflow noise signal power of the i-th car in the specified car production line in the m-th preset time period, represents the electromagnetic interference noise signal power of the i-th car in the specified car production line in the m-th preset time period, It represents the average signal power of the i-th car in the specified car production line in the m-th preset time period, Indicates the preset airflow noise power threshold, Indicates the preset electromagnetic interference noise power threshold, Indicates the preset airflow signal-to-noise ratio threshold. Indicates the preset electromagnetic interference signal-to-noise ratio threshold.

[0047] In this embodiment, the aforementioned database is a database for storing various types of setting data established before the design of a method for constructing a smart product risk prediction model based on machine learning provided in an embodiment of the present application. The database includes but is not limited to airflow noise power, electromagnetic interference power, fault response time, etc., and the various numerical values ​​therein are directly set by technical personnel. For example, the preset airflow noise power threshold is represented by the average value of the airflow noise power of the automobile production line in the historical time period in the database, the preset electromagnetic interference noise power threshold is represented by the average value of the electromagnetic interference noise power of the automobile production line in the historical time period in the database, the preset airflow signal-to-noise ratio threshold is represented by the average value of the airflow signal-to-noise ratio of the automobile production line in the historical time period in the database, and the preset electromagnetic interference signal-to-noise ratio threshold is represented by the average value of the electromagnetic interference signal-to-noise ratio of the automobile production line in the historical time period in the database.

[0048] It should be understood that the algorithm of this embodiment combines the automobile production noise data for comprehensive analysis to obtain the automobile production noise interference value. The automobile production noise data in the algorithm of this embodiment does not exist independently, but is interrelated. In the automobile production process, processes such as stamping, welding, and painting will generate airflow noise. Airflow noise may affect the working state of the monitoring sensor through mechanical vibration and other means, thereby generating electromagnetic interference noise. For example, when high-speed airflow passes through the monitoring sensor, it may cause vibration inside the equipment, which may be converted into electromagnetic noise. Electromagnetic interference may cause the sensor or actuator inside the equipment to work abnormally, thereby changing the state and noise level of the airflow. The interference of airflow noise on data transmission may cause a reduction in the average power of the signal. This reduction may be manifested as a reduction in signal amplitude, frequency offset or phase distortion, etc. The parameters of the algorithm of this embodiment need to consider the impact on the results together.

[0049] Specifically, assuming that the airflow noise power error value The range is 0.6-1, the electromagnetic interference noise power error value The range is 0.6-1, and the error degree of airflow signal-to-noise ratio is The range is 0.6-1, and the electromagnetic interference signal-to-noise ratio error value The range is 0.6-1, as shown in Table 1, which is a statistical table of changes in automobile production noise interference values ​​provided in the embodiment of the present application:

[0050] Table 1 Statistics of changes in noise interference values ​​in automobile production

[0051]

[0052] From the above table, we can see that as the airflow noise power error value , Electromagnetic interference noise power error value , Airflow signal-to-noise ratio error value and electromagnetic interference signal-to-noise ratio error value The noise interference value of automobile production gradually increases, which means that the influence of the noise of the designated automobile production line on the transmission of automobile production data to the preset data monitoring center is gradually increasing, realizing the accurate quantification of the influence of the noise of the designated automobile production line on the transmission of automobile production data to the preset data monitoring center, thereby realizing the improvement of data accuracy for automobile manufacturing risk prediction based on machine learning.

[0053] Furthermore, the specific process of judging whether to perform interference adjustment based on the automobile production noise interference value is as follows: the first step is to judge whether the automobile production noise interference value is not higher than the preset production noise threshold obtained from the database. When the automobile production noise interference value is not higher than the preset production noise threshold, it indicates that the noise impact is qualified and no interference adjustment is performed. Otherwise, the second step is executed; the second step is to send a prompt to the preset personnel to adjust the airflow speed. When the monitored automobile production noise interference value is not higher than the preset production noise threshold, it indicates that the noise impact is qualified and the interference adjustment is stopped. Otherwise, the third step is executed. Adjusting the airflow speed means increasing the length of the ventilation duct on the automobile production line to reduce the noise generated by air flow; the third step is to perform interference shielding. When the monitored automobile production noise interference value is not higher than the preset production noise threshold, it indicates that the noise impact is qualified and the interference adjustment is stopped. Otherwise, an alarm prompt is sent. Interference shielding means reducing electromagnetic interference by using shielded cables and double-shielded twisted pair cables.

[0054] In this embodiment, the preset production noise threshold is represented by the average value of the historical automobile production noise interference value in the database, the length of the ventilation duct is increased in the preset monitoring area on the automobile production line, and shielded cables and double-shielded twisted pairs are installed in the preset monitoring area on the automobile production line. The shielding layer of the shielded cable is usually made of metal mesh, metal foil or metal braid, which can effectively block the interference of external electromagnetic fields on the internal signals of the cable, and also prevent the electromagnetic field generated by the internal signals of the cable from interfering with the outside world. The double-shielded twisted pair has two shielding layers, the outer shielding layer can block the interference of external electromagnetic fields, and the inner shielding layer is used to reduce the crosstalk between the internal signals of the cable; the data accuracy of automobile manufacturing risk prediction based on machine learning is improved.

[0055] Furthermore, the specific process of obtaining the automobile production noise interference value after interference adjustment and combining it with the obtained sensor processing capacity data to obtain the sensor processing capacity evaluation value is as follows: the sum of the actual throughput and the preset throughput threshold obtained from the database is compared with 2 times the preset throughput threshold to obtain the throughput error degree value (that is, the value in the restriction expression of the sensor processing capacity evaluation value). ), the throughput error degree value is used to reflect the deviation of the processing efficiency of the preset data monitoring center for the transmitted automobile production noise data; the sum of the actual response time and the preset response time threshold obtained from the database is compared with 2 times the preset response time threshold to obtain the response time error degree value (that is, the value in the limiting expression of the sensor processing capability evaluation value). ), the response time error degree value reflects the deviation of the processing speed of the preset data monitoring center for the transmitted automobile production noise data; the sensor processing capability evaluation value is obtained by combining the throughput error degree value, the response time error degree value and the automobile production noise interference value after interference adjustment.

[0056] Among them, the limiting expression of the sensor processing capability evaluation value is as follows:

[0057] ;

[0058] ;

[0059] ;

[0060] In the formula, represents the sensor processing capability evaluation value of the i-th car in the specified car production line in the m-th preset time period, , i represents the number of the car on the specified car production line, s represents the total number of cars on the specified car production line, , m represents the number of the preset time period, z represents the total number of preset time periods, It represents the throughput error value of the i-th car in the m-th preset time period of the specified car production line. It represents the response time error value of the i-th car in the specified car production line in the m-th preset time period. It represents the automobile production noise interference value after the interference adjustment of the i-th automobile in the specified automobile production line in the m-th preset time period, represents the actual throughput of the i-th car in the m-th preset time period of the specified car production line, represents the actual response time of the i-th car in the specified car production line in the m-th preset time period, Indicates the preset throughput threshold, represents the preset response time threshold, and e represents a natural constant.

[0061] In this embodiment, the preset throughput threshold is represented by the average value of the throughput monitored on the automobile production line in the historical time period in the database, and the preset response time threshold is represented by the average value of the response time monitored on the automobile production line in the historical time period in the database.

[0062] It should be understood that the algorithm of this embodiment combines the sensor processing capability data and the automobile production noise interference value for comprehensive analysis to obtain the sensor processing capability evaluation value. In the algorithm of this embodiment, the sensor processing capability data and the automobile production noise interference value do not exist independently, but are interrelated. An increase in the automobile production noise interference value may reduce the actual throughput. More noise interference may lead to an increase in the data transmission error rate and an increase in the number of data retransmissions, thereby reducing the amount of data successfully processed per unit time. Noise interference may cause data transmission delays, thereby increasing the actual response time. The parameters of the algorithm of this embodiment need to be considered together and simultaneously in terms of their impact on the results.

[0063] Specifically, assuming that the throughput error value The range is 0.6-1, and the response time error value The range is 0.6-1, and the automobile production noise interference value after interference adjustment The range is 3-4, such as Figure 3 As shown, it is a statistical diagram of the change of the sensor processing capability evaluation value provided in the embodiment of the present application, which is composed of Figure 3 It can be seen that when the response time error value Fixed to 0.6, automobile production noise interference value after interference adjustment When it is fixed at 3, as the throughput error value gradually increases, the sensor processing capability assessment value gradually increases, which means that the sensor's ability to process transmitted data gradually increases, achieving accurate quantification of the processing capability of the preset data monitoring center for the transmitted automobile production noise data in the intelligent automobile manufacturing process, thereby achieving improved data accuracy for automobile manufacturing risk prediction based on machine learning.

[0064] Furthermore, the specific process of judging whether to adjust the sensor capacity based on the sensor processing capacity evaluation value is as follows: Step 1, judging whether the sensor processing capacity evaluation value is not lower than the preset processing capacity threshold obtained from the database. When the sensor processing capacity evaluation value is not lower than the preset processing capacity threshold, it indicates that the sensor processing capacity is qualified and the sensor capacity adjustment is not performed. Otherwise, step 2 is executed; Step 2, multi-threaded transmission is performed. When the monitored sensor processing capacity evaluation value is not lower than the preset processing capacity threshold, it indicates that the sensor processing capacity is qualified and the sensor capacity adjustment is stopped. Otherwise, step 3 is executed. Multi-threaded transmission is used to speed up data transmission; Step 3, resource consumption optimization is performed. When the monitored sensor processing capacity evaluation value is not lower than the preset processing capacity threshold, it indicates that the sensor processing capacity is qualified and the sensor capacity adjustment is stopped. Otherwise, an alarm prompt is sent. Resource consumption optimization means reducing computing and storage resource consumption through lightweight encryption algorithms.

[0065] In this embodiment, the preset processing capacity threshold is represented by the average value of the historical sensor processing capacity evaluation value in the database. The lightweight encryption algorithm is an encryption algorithm designed for resource-constrained environments (such as Internet of Things devices, wireless sensor networks, etc.), which aims to ensure security while meeting the restrictions on processing capacity, storage space and energy consumption. This embodiment reduces the amount of calculation and resource consumption through the PRESENT algorithm while resisting various cryptographic attacks, uses a multi-threaded library or framework in a programming language (such as Java's Thread class) to achieve multi-threaded transmission, and uses network bandwidth and CPU resources to achieve high-speed data transmission; it achieves the improvement of data accuracy for automobile manufacturing risk prediction based on machine learning.

[0066] Furthermore, the specific process of obtaining the automobile risk assessment value based on the acquired automobile production abnormal status data and the sensor processing capability assessment value after the sensor capability adjustment is as follows: the total automobile repair time and the preset total automobile repair time threshold obtained from the database are ratioed to obtain the repair time error degree value (that is, the error degree value in the restriction expression of the automobile risk assessment value). ), the repair time error value is used to reflect the deviation of the fault repair time of a specified automobile production line; the total number of automobile failures and the preset total number of automobile failures threshold obtained from the database are used to perform a ratio operation to obtain the fault number error value (i.e., the value in the restriction expression of the automobile risk assessment value). ), the fault number error degree value is used to reflect the deviation of the fault repair times of a specified automobile production line; the automobile risk assessment value is obtained by combining the repair time error degree value, the fault number error degree value and the sensor processing capability assessment value after sensor capability adjustment.

[0067] Among them, the restricted expression of automobile risk assessment value is as follows:

[0068] ;

[0069] ;

[0070] ;

[0071] In the formula, represents the automobile risk assessment value of a specified automobile production line in the mth preset time period, , m represents the number of the preset time period, z represents the total number of preset time periods, It represents the repair time error value of the specified automobile production line in the mth preset time period. It represents the error value of the number of failures in the mth preset time period of the specified automobile production line. represents the sensor processing capability evaluation value of the i-th car in the specified car production line after the sensor capability adjustment in the m-th preset time period, It represents the total repair time of the specified automobile production line in the mth preset time period. It represents the total number of automobile failures in the mth preset time period of a specified automobile production line. Indicates the preset total vehicle repair time threshold, represents the preset threshold of the total number of automobile failures, and e represents a natural constant.

[0072] In this embodiment, the preset total automobile repair time threshold is represented by the average value of total automobile repairs on the automobile production line in the historical time period in the database, and the preset total automobile failure number threshold is represented by the average value of total automobile failure number on the automobile production line in the historical time period in the database.

[0073] It should be understood that the algorithm of this embodiment combines the automobile production abnormal state data and the sensor processing capability evaluation value for comprehensive analysis to obtain the automobile risk evaluation value. In the algorithm of this embodiment, the automobile production abnormal state data and the sensor processing capability evaluation value do not exist independently, but are interrelated. A reduction in the sensor processing capability evaluation value may lead to an increase in the total repair time of the automobile. When the sensor processing capacity is insufficient, it may not be possible to detect and diagnose faults in automobile production in a timely and accurate manner, thereby extending the time for fault repair. An increase in the number of faults usually leads to an extension of the repair time, and an increase in the sensor processing capability evaluation value helps to reduce the total number of automobile faults. The more accurate the data monitoring, the earlier the sensor can detect potential faults, thereby reducing the number of faults. The parameters of the algorithm of this embodiment need to jointly consider the impact on the results at the same time, so as to achieve the accurate quantification of the abnormal situation of automobile production in the process of evaluating the intelligent automobile manufacturing, and thus achieve the improvement of the data accuracy of automobile manufacturing risk prediction based on machine learning.

[0074] Furthermore, the specific process of risk prediction classification based on the constructed automobile risk prediction model is as follows: determine whether the automobile risk assessment value is not higher than the preset risk threshold obtained from the database. When the automobile risk assessment value is not higher than the preset risk threshold, obtain the corresponding automobile risk assessment value, otherwise it is directly marked as high risk; input the obtained automobile risk assessment value that is not higher than the preset risk threshold and the corresponding automobile production noise interference value after interference adjustment and the sensor processing capability assessment value after sensor capability adjustment into the constructed automobile risk prediction model to obtain the risk prediction classification result.

[0075] In this embodiment, the preset risk threshold is represented by the average value of historical automobile risk assessment values ​​in the database. The automobile risk prediction model can learn the risk characteristics and patterns of the production line, warn of potential production risks in advance, and provide support for production decisions. Through risk prediction classification, the resource allocation and scheduling of the production line can be optimized to improve production efficiency and quality; the data accuracy of automobile manufacturing risk prediction based on machine learning is improved.

[0076] To summarize, the embodiment of the present application obtains a sensor processing capability evaluation value by combining the acquired sensor processing capability data and the automobile production noise interference value after interference adjustment and determines whether to adjust the sensor capability, and then obtains an automobile risk assessment value by combining the automobile production abnormal status data and the sensor processing capability evaluation value after sensor capability adjustment. Finally, risk prediction classification is performed based on the constructed automobile risk prediction model, thereby achieving an improvement in the accuracy of constructing the automobile risk prediction model, and further achieving an improvement in the data accuracy of automobile manufacturing risk prediction based on machine learning, effectively solving the problem of inaccurate data for automobile manufacturing risk prediction based on machine learning in the prior art.

[0077] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0078] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0079] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0080] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0081] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0082] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for constructing an intelligent product risk prediction model based on machine learning, characterized in that: The following steps are involved: S1, obtaining an automobile production noise interference value according to the automobile production noise data of the specified automobile production line, wherein the automobile production noise interference value is used to evaluate the influence of the noise of the specified automobile production line on the transmission of automobile production data to a preset data monitoring center; S2, judging whether to perform interference adjustment based on the automobile production noise interference value, obtaining the automobile production noise interference value after the interference adjustment and combining it with the obtained sensor processing capability data to obtain a sensor processing capability evaluation value, wherein the sensor processing capability evaluation value is used to evaluate the processing capability of the preset data monitoring center for the transmitted automobile production noise data in the intelligent automobile manufacturing process; S3, judging whether to adjust the sensor capacity based on the sensor processing capacity evaluation value, obtaining a vehicle risk evaluation value based on the acquired abnormal state data of automobile production and the sensor processing capacity evaluation value after the sensor capacity adjustment, wherein the automobile risk evaluation value is used to evaluate abnormal automobile production in the intelligent automobile manufacturing process; S4, performing risk prediction classification based on the constructed automobile risk prediction model, wherein the risk prediction classification is used to reflect the risk level of a specified automobile production line, and the automobile risk prediction model represents a model constructed by a machine learning algorithm for outputting risk prediction classification results; The specific process of obtaining the automobile production noise interference value according to the automobile production noise data of the specified automobile production line is as follows: The airflow noise power error degree value is obtained by performing a ratio operation on the sum of the airflow noise signal power and the preset airflow noise power threshold obtained from the database and twice the preset airflow noise power threshold; The electromagnetic interference noise signal power and the preset electromagnetic interference noise power threshold obtained from the database are ratioed to twice the preset electromagnetic interference noise power threshold to obtain an electromagnetic interference noise power error degree value; The airflow signal-to-noise ratio is obtained by performing a ratio operation on the signal average power and the airflow noise signal power, and the electromagnetic signal-to-noise ratio is obtained by performing a ratio operation on the magnetic interference noise signal power and the electric signal average power; Performing a ratio operation on twice the preset airflow signal-to-noise ratio threshold obtained from the database and the sum of the airflow signal-to-noise ratio and the preset airflow signal-to-noise ratio threshold to obtain an airflow signal-to-noise ratio error degree value; Performing a ratio operation on the sum of the electromagnetic signal-to-noise ratio and the preset electromagnetic interference signal-to-noise ratio threshold value obtained from the database and twice the preset electromagnetic interference signal-to-noise ratio threshold value to obtain an electromagnetic interference signal-to-noise ratio error degree value; The automobile production noise interference value is obtained by combining the airflow noise power error value, the electromagnetic interference noise power error value, the airflow signal-to-noise ratio error value and the electromagnetic interference signal-to-noise ratio error value; The limiting expression of the automobile production noise interference value is as follows: ； ； ； ； ； In the formula, It represents the automobile production noise interference value of the i-th automobile in the specified automobile production line in the m-th preset time period, , i represents the number of the car on the specified car production line, s represents the total number of cars on the specified car production line, , m represents the number of the preset time period, z represents the total number of preset time periods, It represents the airflow noise power error value of the i-th car in the specified car production line in the m-th preset time period, It represents the electromagnetic interference noise power error value of the i-th car in the specified car production line in the m-th preset time period, It represents the error degree of airflow signal-to-noise ratio of the i-th car in the specified car production line in the m-th preset time period. It represents the electromagnetic interference signal-to-noise ratio error value of the i-th car in the specified car production line in the m-th preset time period, represents the airflow noise signal power of the i-th car in the specified car production line in the m-th preset time period, represents the electromagnetic interference noise signal power of the i-th car in the specified car production line in the m-th preset time period, It represents the average signal power of the i-th car in the specified car production line in the m-th preset time period, Indicates the preset airflow noise power threshold, Indicates the preset electromagnetic interference noise power threshold, Indicates the preset airflow signal-to-noise ratio threshold. Indicates the preset electromagnetic interference signal-to-noise ratio threshold; The specific process of judging whether to perform interference adjustment based on the automobile production noise interference value is as follows: The first step is to determine whether the automobile production noise interference value is not higher than the preset production noise threshold obtained from the database. When the automobile production noise interference value is not higher than the preset production noise threshold, no interference adjustment is performed, otherwise the second step is executed; The second step is to send a reminder to the preset personnel to adjust the airflow speed. When the monitored automobile production noise interference value is not higher than the preset production noise threshold, the interference adjustment is stopped, otherwise the third step is executed; The third step is to perform interference shielding. When the monitored automobile production noise interference value is not higher than the preset production noise threshold, the interference adjustment is stopped. Otherwise, an alarm prompt is sent; The specific process of obtaining the automobile production noise interference value after interference adjustment and combining it with the obtained sensor processing capability data to obtain the sensor processing capability evaluation value is as follows: A throughput error degree value is obtained by performing a ratio operation on the sum of the actual throughput and the preset throughput threshold obtained from the database and twice the preset throughput threshold, wherein the throughput error degree value is used to reflect the deviation of the processing efficiency of the preset data monitoring center for the transmitted automobile production noise data; The sum of the actual response time and the preset response time threshold obtained from the database is ratioed to twice the preset response time threshold to obtain a response time error value, wherein the response time error value reflects the deviation of the processing speed of the transmitted automobile production noise data by the preset data monitoring center; The sensor processing capability evaluation value is obtained by combining the throughput error degree value, the response time error degree value and the automobile production noise interference value after interference adjustment; The specific process of judging whether to adjust the sensor capability based on the sensor processing capability evaluation value is as follows: Step 1: determine whether the sensor processing capability evaluation value is not lower than the preset processing capability threshold obtained from the database. When the sensor processing capability evaluation value is not lower than the preset processing capability threshold, the sensor capability is not adjusted. Otherwise, step 2 is executed. Step 2: Perform multi-threaded transmission. When the monitored sensor processing capacity evaluation value is not lower than the preset processing capacity threshold, stop adjusting the sensor capacity. Otherwise, execute step 3. Step three, optimize resource consumption. When the monitored sensor processing capacity evaluation value is not lower than the preset processing capacity threshold, stop adjusting the sensor capacity. Otherwise, send an alarm prompt.

2. The method for constructing a smart product risk prediction model based on machine learning as claimed in claim 1, characterized in that: The automobile production noise data includes airflow noise signal power, electromagnetic interference noise signal power and signal average power; The sensor processing capability data includes actual throughput and actual response time; The actual throughput represents the amount of data corresponding to the automobile production data processed and transmitted by the preset data monitoring center per unit time; The actual response time represents the time for the automobile production data to be transmitted to the preset data monitoring center; The automobile production abnormal status data includes the total automobile repair time and the total number of automobile failures.

3. The method for constructing a smart product risk prediction model based on machine learning as claimed in claim 2, characterized in that: The automobile production noise data is obtained by measuring the noise conditions of a designated automobile production line within a preset time period using a noise acquisition device; The noise acquisition equipment includes a sound level meter, an electromagnetic interference tester and a power meter; The sensor processing capability data is obtained by measuring the sensor processing conditions of a specified automobile production line within a preset time period by a capability processing device; The capability processing equipment includes a network data flow analyzer and a response time tester; The automobile production abnormal state data is obtained by measuring the abnormal state of automobiles on a designated automobile production line within a preset time period through an abnormality acquisition device; The abnormality acquisition device includes a timer and a counter.

4. The method for constructing a smart product risk prediction model based on machine learning as claimed in claim 1, characterized in that: The specific process of obtaining the automobile risk assessment value based on the acquired automobile production abnormal status data and the sensor processing capability assessment value after the sensor capability adjustment is as follows: Performing a ratio operation on the total automobile repair time and a preset total automobile repair time threshold obtained from a database to obtain a repair time error degree value, wherein the repair time error degree value is used to reflect the deviation of the fault repair time of a specified automobile production line; Performing a ratio operation on the total number of automobile failures and a preset total number of automobile failures threshold obtained from a database to obtain a failure number error degree value, wherein the failure number error degree value is used to reflect the deviation of the number of failure repairs of a specified automobile production line; The automobile risk assessment value is obtained by combining the repair time error degree value, the failure number error degree value and the sensor processing capability assessment value after sensor capability adjustment.

5. The method for constructing a smart product risk prediction model based on machine learning as claimed in claim 1, characterized in that: The specific process of risk prediction classification based on the constructed automobile risk prediction model is as follows: Determine whether the automobile risk assessment value is not higher than the preset risk threshold obtained from the database. When the automobile risk assessment value is not higher than the preset risk threshold, obtain the corresponding automobile risk assessment value. Otherwise, directly mark it as high risk. The obtained automobile risk assessment value that is not higher than the preset risk threshold and the corresponding automobile production noise interference value after interference adjustment and the sensor processing capability assessment value after sensor capability adjustment are input into the constructed automobile risk prediction model to obtain the risk prediction classification result.

Images