An intelligent product quality and safety risk early warning method and system based on big data
By establishing a risk prediction model based on nuclear ridge regression and adaptive random forests, the problems of large-scale data processing and risk prediction in intelligent product quality management are solved, efficient and accurate risk identification and early warning are achieved, and product reliability is improved. Transparency in corporate decision-making.
Patent Information
- Application Number
- CN202410726110.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-06-06
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2044-06-06
AI Technical Summary
The prior art is difficult to efficiently process large-scale data in intelligent product quality management, accurately predict quality risks, and interpret model prediction results, limiting its application in actual production environments.
By obtaining historical data of smart products, including performance data, user behavior data, product life cycle and maintenance data, calculate deviation data and correlation, establish a risk prediction model based on nuclear ridge regression and adaptive random forest models, and conduct risk identification and early warning .
It realizes efficient and accurate identification of potential risks of smart products, monitors product performance and user usage habits in real time, promptly warns of quality crises, and has good interpretability, enhancing the transparency and credibility of corporate decisions.
Smart Images

Figure CN118485307B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of intelligent product quality risk early warning, and in particular to an intelligent product quality safety risk early warning method and system based on big data. Background Art
[0002] In the field of intelligent product quality management, with the development of big data technology, more and more companies are beginning to use data analysis to improve product quality and user experience. Traditional quality control methods mostly rely on manual inspection and empirical judgment, which are not only inefficient but also difficult to cope with large-scale and diverse data. In recent years, some companies have tried to introduce machine learning algorithms to analyze product performance data, user feedback and market information in order to achieve automated quality monitoring and risk warning. However, existing technologies still face challenges in computational efficiency and accuracy when processing large-scale data sets, and are insufficient in explaining model prediction results, which limits their widespread application in actual production environments. Therefore, developing an intelligent product quality risk warning method and system that can efficiently process big data, accurately predict quality risks and have good interpretability is of great significance to improving corporate competitiveness and product reliability. Summary of the invention
[0003] The purpose of the present invention is to provide an intelligent product quality and safety risk early warning method based on big data.
[0004] To achieve the above object, the present invention is implemented according to the following technical solutions:
[0005] The present invention comprises the following steps:
[0006] Obtain historical data of smart products, including performance data, user behavior data, product life cycle and maintenance data, and collect standard data;
[0007] Comparing the performance data with the standard data to obtain deviation data, and calculating the correlation between the user behavior data and the maintenance data to obtain impact data;
[0008] Performing risk identification on the deviation data based on the maintenance data to obtain a first weight of the influencing data, and obtaining a dynamic weight based on the product life cycle and the first weight;
[0009] A risk prediction model is established based on the impact data, the optimized dynamic weight and the maintenance data, the data of the smart product to be detected is input into the risk prediction model to output the risk prediction result, and a risk warning is performed.
[0010] Furthermore, historical data of the smart product is obtained, wherein the historical data includes performance data, user behavior data, product life cycle and maintenance data. The method for collecting standard data includes:
[0011] Performance data includes battery life, response time and functional indicators. User behavior data includes usage frequency, usage patterns and abnormal behaviors that cause smart product failures reported by users. Product life cycle includes service life and update records. Update records include the number of times the software and hardware of smart products have been updated. Maintenance data includes the number of repairs of smart products and the corresponding failure types. Standard data is the average initial performance data of smart products.
[0012] Furthermore, the method of comparing the performance data with the standard data to obtain deviation data includes:
[0013] Box plot analysis uses five numbers to summarize the data distribution. The box plot of performance data is compared with that of standard data. The line extending from the first quartile to the minimum non-outlier value and from the third quartile to the maximum non-outlier value is the box plot whisker. The performance data exceeding the whisker of the box plot is regarded as deviant data, and the deviation value of the deviant data relative to the critical value of the standard data is calculated as the deviation data.
[0014] Furthermore, the method of calculating the correlation between the user behavior data and the maintenance data to obtain the influence data includes:
[0015] Encode user behavior data and convert it into numerical data, convert maintenance data into numerical vectors, and use user data as input data to predict maintenance data:
[0016] Decision function:
[0017]
[0018] in, is the i-th input data, n is the number of input data, is the decision function of the kernel ridge regression model, is the Gaussian kernel function, α i is the kernel ridge regression model corresponding to each training sample The coefficient of , b is the bias term,
[0019] Output of the kernel ridge regression model integrated in the adaptive random forest model:
[0020]
[0021] in, is the prediction output of the adaptive random forest model, α j′ is the coefficient of the jth tree in the random forest, is the base model prediction of the jth tree in the random forest, ∈ j is the error term of the jth tree in the random forest, ρ j (R) is the adjustment factor based on the correlation matrix R, which represents the cosine similarity between trees. The output of the integrated model is:
[0022]
[0023] in, is the prediction data of the integrated model, λ k is the k-th output coefficient of kernel ridge regression and adaptive random forest, K is the number of kernel ridge regression models used in the integrated model, h(·) is ReLU, is the nonlinear activation function,
[0024] Loss function:
[0025]
[0026] in, is the i-th maintenance data, is the i-th predicted value of the ensemble model, λ KRR is the regularization parameter of kernel ridge regression, α i is the coefficient in kernel ridge regression, λ aRF is the regularization parameter of random forest, α j ′ and g j (x) are the coefficients and base model predictions of the jth tree in the random forest, is the variance term, γ is the weight parameter controlling the variance term,
[0027] Check the contribution of input data to the ensemble model:
[0028]
[0029] Where v(S) represents the SHAP value of the subset S of the input data, and v(S∪{i}) represents the SHAP value of the new subset after adding the input data i.
[0030] The input data with SHAP values less than the preset threshold are removed, the filtered input data are input into the integrated model, and the predicted data of the integrated model are used as the influencing data.
[0031] Furthermore, the method of performing risk identification on the deviation data based on the maintenance data to obtain the first weight of the influencing data includes:
[0032] The maintenance data and deviation data are used to establish a data matrix. The maintenance data and deviation data in the data matrix correspond to each other one by one, and the vacant values are filled with 0. The deviation data is used to establish data points, and the data points are clustered:
[0033] Neighborhood:
[0034] N ∈,w(p) ={q∈D|dist(p,q)≤∈∧q≠p∧w(q)≥w min}
[0035] Where D is the set of data points, w(q) is the local density weight of data point q, ∈ is the radius of the neighborhood, and w min is the weight threshold of the neighborhood, dist(p,q) represents the Euclidean distance between data point p and data point q,
[0036] The core points of the neighborhood:
[0037]
[0038] Among them, N ∈,w(p) is the number of data points in the neighborhood, MinPts is the minimum point threshold used to determine whether a data point can become a core point, ν(p) is the noise term, α is the parameter for adjusting the noise term, and ν max is the maximum threshold of the noise term, is the local anomaly factor of the data point,
[0039] Cluster Center Update:
[0040]
[0041] Among them, x is the data point in cluster j, C j is cluster j, I j is the influence term of cluster j, is the variance of the data points in cluster j,
[0042] Objective function:
[0043]
[0044] Among them, β and γ are weight parameters, K is the number of clusters, λ is the weight of the regularization term, and F j is the compactness of cluster j, N noise is the distribution of noise points, R is the global regularization term, is the L1 norm,
[0045] The intra-cluster failure rate is calculated according to the maintenance data corresponding to the deviation data in the clustering result in the data matrix as the first weight.
[0046] Furthermore, a method for obtaining a dynamic weight based on the product life cycle and the first weight includes:
[0047] The product life cycle is normalized and preprocessed. The average historical maintenance rate during the product life cycle is calculated through maintenance data based on the exponential decay model as the second weight affecting the data. The average of the first weight and the second weight is calculated as the dynamic weight.
[0048] Furthermore, a method for establishing a risk prediction model according to the impact data, the optimized dynamic weight and the maintenance data includes:
[0049] The weighted influence data is obtained by multiplying the influence data and the dynamic weight. The weighted influence data and the maintenance data are used to establish a data set. The data set is divided into a training set and a test set. The maintenance data is predicted based on the weighted influence data, and a regression model is established:
[0050] Expected prediction results:
[0051]
[0052] Among them, Φ represents the probability distribution of the prediction results, is the predictor variable, are input features, represents the training set, is the set of model parameters of the regression model, Λ is the variational distribution of the model parameters of the regression model,
[0053] Goodness of fit:
[0054]
[0055] Where Ξ represents the goodness of fit, ω is the parameter that adjusts the weight between the variational distribution and the likelihood function, ν is the model parameter, i is the index of ν, ρ is the variational distribution, Υ is the uniform prior distribution, Π is the likelihood function, ∈ is the uncertainty parameter,
[0056] Probability density function:
[0057]
[0058] Among them, τ n is the noise term, is the predicted value of the regression model, α is the adjustment function,
[0059] Goodness of fit for prediction uncertainty:
[0060]
[0061] Use the test set to test the accuracy of the regression model.
[0062] An intelligent product quality risk early warning system based on big data, characterized by comprising:
[0063] The acquisition module acquires historical data of smart products, including performance data, user behavior data, product life cycle and maintenance data, and collects standard data;
[0064] an optimization module, performing risk identification on the deviation data based on the maintenance data to obtain a first weight of the influencing data, and obtaining a dynamic weight based on the product life cycle and the first weight;
[0065] a calculation module, comparing the performance data with the standard data to obtain deviation data, calculating the correlation degree of the user behavior data with the maintenance data to obtain influence data, and establishing a risk prediction model according to the influence data, the optimized dynamic weight and the maintenance data;
[0066] The output module inputs the data of the smart product to be tested into the risk prediction model to output the risk prediction result and issue a risk warning.
[0067] In a second aspect, an embodiment of the present application further provides an electronic device, including:
[0068] A processor; and a memory arranged to store computer executable instructions, which when executed cause the processor to perform the method steps described in the first aspect.
[0069] In a third aspect, an embodiment of the present application further provides a computer-readable storage medium, which stores one or more programs. When the one or more programs are executed by an electronic device including multiple applications, the electronic device executes the method steps described in the first aspect.
[0070] The beneficial effects of the present invention are:
[0071] The present invention is an intelligent product quality safety risk early warning method and system based on big data. Compared with the prior art, the present invention has the following technical effects:
[0072] The present invention achieves efficient and accurate identification of potential risks of smart products by comprehensively utilizing big data technology. The present invention can monitor product performance in real time, timely capture user usage habits and abnormal changes in the product life cycle, and thus quickly respond to and warn of possible quality crises. In addition, the present invention also has good interpretability, allowing enterprises to clearly understand the basis for risk warnings, and enhancing the transparency and credibility of decision-making. BRIEF DESCRIPTION OF THE DRAWINGS
[0073] Figure 1 This is a flowchart of the steps of a smart product quality and safety risk early warning method and system based on big data of the present invention;
[0074] Figure 2It is a schematic diagram of the structure of an electronic device in an embodiment of this specification. DETAILED DESCRIPTION
[0075] The present invention is further described below by means of specific embodiments. The illustrative embodiments and descriptions of the present invention are used to explain the present invention but are not intended to limit the present invention.
[0076] The present invention provides a smart product quality safety risk early warning method based on big data, comprising the following steps:
[0077] like Figure 1 As shown, in this embodiment, the following steps are included:
[0078] Obtain historical data of smart products, including performance data, user behavior data, product life cycle and maintenance data, and collect standard data;
[0079] In the actual evaluation, in a smart bracelet enterprise, the data obtained include:
[0080]
[0081] Among them, the accuracy of heart rate monitoring is a functional indicator, and the standard data is: battery life: 10 hours, response time: 150ms, functional indicators: heart rate monitoring accuracy 90%;
[0082] Comparing the performance data with the standard data to obtain deviation data, and calculating the correlation between the user behavior data and the maintenance data to obtain impact data;
[0083] In actual evaluation, the deviation data:
[0084] User ID Battery life (hours) Response time (ms) Heart rate monitoring accuracy (%) 1 -2 50 -15.00% 2 2 0 -10.00% 3 -3 30 -20.00% 4 1 10 -2.00%
[0085] Impact data: User 1: 0.7 screen failure, 0.3 battery exhausted, User 2: 0.4 screen failure, 0.8 battery exhausted, User 3: 0.95 screen failure, 0.4 battery exhausted, User 4: 0.9 screen failure, 0.8 battery exhausted;
[0086] Performing risk identification on the deviation data based on the maintenance data to obtain a first weight of the influencing data, and obtaining a dynamic weight based on the product life cycle and the first weight;
[0087] In the actual evaluation, the first weight is: User 1: 0.5 screen failure, 0 battery exhaustion, User 2: 1 screen failure, 0.5 battery exhaustion, User 3: 0.5 screen failure, 0 battery exhaustion, User 4: 1 screen failure, 0.5 battery exhaustion, and the dynamic weight is: User 1: 0.6 screen failure, 0.15 battery exhaustion, User 2: 0.7 screen failure, 0.7 battery exhaustion, User 3: 0.73 screen failure, 0.2 battery exhaustion, User 4: 0.8 screen failure, 0.8 battery exhaustion;
[0088] A risk prediction model is established based on the impact data, the optimized dynamic weight and the maintenance data, the data of the smart product to be detected is input into the risk prediction model to output the risk prediction result, and a risk warning is performed.
[0089] In this embodiment, historical data of smart products is obtained, and the historical data includes performance data, user behavior data, product life cycle and maintenance data. The method of collecting standard data includes:
[0090] Performance data includes battery life, response time and functional indicators. User behavior data includes usage frequency, usage patterns and abnormal behaviors that cause smart product failures reported by users. Product life cycle includes service life and update records. Update records include the number of times the software and hardware of smart products have been updated. Maintenance data includes the number of repairs of smart products and the corresponding failure types. Standard data is the average initial performance data of smart products.
[0091] In this embodiment, the method of comparing the performance data with the standard data to obtain deviation data includes:
[0092] Box plot analysis uses five numbers to summarize the data distribution. The box plot of performance data is compared with that of standard data. The line extending from the first quartile to the minimum non-outlier value and from the third quartile to the maximum non-outlier value is the box plot whisker. The performance data exceeding the whisker of the box plot is regarded as deviant data, and the deviation value of the deviant data relative to the critical value of the standard data is calculated as the deviation data.
[0093] In this embodiment, the method of calculating the correlation degree of the user behavior data to the maintenance data to obtain the influence data includes:
[0094] Encode user behavior data and convert it into numerical data, convert maintenance data into numerical vectors, and use user data as input data to predict maintenance data:
[0095] Decision function:
[0096]
[0097] in, is the i-th input data, n is the number of input data, is the decision function of the kernel ridge regression model, is the Gaussian kernel function, α i is the kernel ridge regression model corresponding to each training sample The coefficient of , b is the bias term,
[0098] Output of the kernel ridge regression model integrated in the adaptive random forest model:
[0099]
[0100] in, is the prediction output of the adaptive random forest model, α j ′ is the coefficient of the jth tree in the random forest, is the base model prediction of the jth tree in the random forest, ∈ j is the error term of the jth tree in the random forest, ρ j (R) is the adjustment factor based on the correlation matrix R, which represents the cosine similarity between trees. The output of the integrated model is:
[0101]
[0102] in, is the prediction data of the integrated model, λ k is the k-th output coefficient of kernel ridge regression and adaptive random forest, K is the number of kernel ridge regression models used in the integrated model, h(·) is ReLU, is the nonlinear activation function,
[0103] Loss function:
[0104]
[0105] in, is the i-th maintenance data, is the i-th predicted value of the ensemble model, λ KRR is the regularization parameter of kernel ridge regression, α i is the coefficient in kernel ridge regression, λ aRF is the regularization parameter of random forest, α j ′ and g j (x) are the coefficients and base model predictions of the jth tree in the random forest, is the variance term, γ is the weight parameter controlling the variance term,
[0106] Check the contribution of input data to the ensemble model:
[0107]
[0108] Where v(S) represents the SHAP value of the subset S of the input data, and v(S∪{i}) represents the SHAP value of the new subset after adding the input data i.
[0109] The input data with SHAP values less than the preset threshold are removed, the filtered input data are input into the integrated model, and the predicted data of the integrated model are used as the influencing data.
[0110] In this embodiment, the method for performing risk identification on the deviation data based on the maintenance data to obtain the first weight of the influencing data includes:
[0111] The maintenance data and deviation data are used to establish a data matrix. The maintenance data and deviation data in the data matrix correspond to each other one by one, and the vacant values are filled with 0. The deviation data is used to establish data points, and the data points are clustered:
[0112] Neighborhood:
[0113] N ∈,w(p) ={q∈D|dist(p,q)≤∈∧q≠p∧w(q)≥w min}
[0114] Where D is the set of data points, w(q) is the local density weight of data point q, ∈ is the radius of the neighborhood, and w min is the weight threshold of the neighborhood, dist(p,q) represents the Euclidean distance between data point p and data point q,
[0115] The core points of the neighborhood:
[0116]
[0117] Among them, N ∈,w(p) is the number of data points in the neighborhood, MinPts is the minimum point threshold used to determine whether a data point can become a core point, ν(p) is the noise term, α is the parameter for adjusting the noise term, and ν max is the maximum threshold of the noise term, is the local anomaly factor of the data point,
[0118] Cluster Center Update:
[0119]
[0120] Among them, x is the data point in cluster j, C j is cluster j, I j is the influence term of cluster j, is the variance of the data points in cluster j,
[0121] Objective function:
[0122]
[0123] Among them, β and γ are weight parameters, K is the number of clusters, λ is the weight of the regularization term, and F j is the compactness of cluster j, N noise is the distribution of noise points, R is the global regularization term, is the L1 norm,
[0124] The intra-cluster failure rate is calculated according to the maintenance data corresponding to the deviation data in the clustering result in the data matrix as the first weight.
[0125] In this embodiment, the method for obtaining a dynamic weight based on the product life cycle and the first weight includes:
[0126] The product life cycle is normalized and preprocessed. The average historical maintenance rate during the product life cycle is calculated through maintenance data based on the exponential decay model as the second weight affecting the data. The average of the first weight and the second weight is calculated as the dynamic weight.
[0127] In this embodiment, the method for establishing a risk prediction model according to the impact data, the optimized dynamic weight and the maintenance data includes:
[0128] The weighted influence data is obtained by multiplying the influence data and the dynamic weight. The weighted influence data and the maintenance data are used to establish a data set. The data set is divided into a training set and a test set. The maintenance data is predicted based on the weighted influence data, and a regression model is established:
[0129] Expected prediction results:
[0130]
[0131] Among them, Φ represents the probability distribution of the prediction results, is the predictor variable, are input features, represents the training set, is the set of model parameters of the regression model, Λ is the variational distribution of the model parameters of the regression model,
[0132] Goodness of fit:
[0133]
[0134] Where Ξ represents the goodness of fit, ω is the parameter that adjusts the weight between the variational distribution and the likelihood function, ν is the model parameter, i is the index of ν, ρ is the variational distribution, Υ is the uniform prior distribution, Π is the likelihood function, ∈ is the uncertainty parameter,
[0135] Probability density function:
[0136]
[0137] Among them, τn is the noise term, is the predicted value of the regression model, α is the adjustment function,
[0138] Goodness of fit for prediction uncertainty:
[0139]
[0140] Use the test set to test the accuracy of the regression model.
[0141] An intelligent product quality risk early warning system based on big data, characterized by comprising:
[0142] The acquisition module acquires historical data of smart products, including performance data, user behavior data, product life cycle and maintenance data, and collects standard data;
[0143] an optimization module, performing risk identification on the deviation data based on the maintenance data to obtain a first weight of the influencing data, and obtaining a dynamic weight based on the product life cycle and the first weight;
[0144] a calculation module, comparing the performance data with the standard data to obtain deviation data, calculating the correlation degree of the user behavior data with the maintenance data to obtain influence data, and establishing a risk prediction model according to the influence data, the optimized dynamic weight and the maintenance data;
[0145] The output module inputs the data of the smart product to be tested into the risk prediction model to output the risk prediction result and issue a risk warning.
[0146] Figure 2 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present application. Figure 2 At the hardware level, the electronic device includes a processor, and optionally also includes an internal bus, a network interface, and a memory. The memory may include a memory, such as a high-speed random access memory (RAM), and may also include a non-volatile memory (non-volatile memory), such as at least one disk storage. Of course, the electronic device may also include hardware required for other services.
[0147] The processor, network interface and memory can be interconnected through an internal bus, which can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 2 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or only one type of bus.
[0148] The memory is used to store the program. Specifically, the program may include a program code, and the program code includes a computer operation instruction. The memory may include a memory and a non-volatile memory, and provides instructions and data to the processor.
[0149] The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it, forming an intelligent product quality risk early warning device based on big data at the logical level. The processor executes the program stored in the memory and is specifically used to execute any of the aforementioned intelligent product quality safety risk early warning methods and systems based on big data.
[0150] The above application Figure 1The embodiment shown discloses a method and system for early warning of intelligent product quality and safety risks based on big data, which can be applied to a processor or implemented by a processor. The processor may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by an integrated logic circuit of hardware in the processor or instructions in the form of software. The above processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present application can be implemented or executed. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in the embodiment of the present application can be directly embodied as a hardware decoding processor for execution, or a combination of hardware and software modules in the decoding processor for execution. The software module can be located in a storage medium mature in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.
[0151] The electronic device may also perform Figure 1 An intelligent product quality and safety risk early warning method and system based on big data is implemented Figure 1 The functions of the illustrated embodiment will not be described in detail in the embodiments of the present application.
[0152] An embodiment of the present application also proposes a computer-readable storage medium, which stores one or more programs, and the one or more programs include instructions. When the instructions are executed by an electronic device including multiple applications, any one of the aforementioned big data-based intelligent product quality and safety risk warning methods and systems is executed.
[0153] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.
[0154] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes 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 generate 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.
[0155] 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.
[0156] 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. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0157] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.
[0158] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.
[0159] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.
[0160] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.
[0161] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment or an embodiment in combination with software and hardware. Moreover, the present application may adopt 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.) that contain computer-usable program code.
[0162] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims
1. An intelligent product quality and safety risk early warning method based on big data, characterized in that: The following steps are involved: Obtain historical data and standard data of smart products, wherein the historical data includes performance data, user behavior data, product life cycle and maintenance data; The performance data is compared with the standard data to obtain deviation data, and the correlation between the user behavior data and the maintenance data is calculated to obtain impact data. Encode user behavior data and convert it into numerical data, convert maintenance data into numerical vectors, and use user data as input data to predict maintenance data: Decision function: in, is the i-th input data, n is the number of input data, is the decision function of the kernel ridge regression model, is the Gaussian kernel function, α i is the kernel ridge regression model corresponding to each training sample The coefficient of , b is the bias term, Output of the kernel ridge regression model integrated in the adaptive random forest model: in, is the prediction output of the adaptive random forest model, α j ′ is the coefficient of the jth tree in the random forest, is the base model prediction of the jth tree in the random forest, ∈ j is the error term of the jth tree in the random forest, ρ j (R) is the adjustment factor based on the correlation matrix R, which represents the cosine similarity between trees. Output of the ensemble model: in, is the prediction data of the integrated model, λ k is the k-th output coefficient of kernel ridge regression and adaptive random forest, K is the number of kernel ridge regression models used in the integrated model, h(·) is ReLU, is the nonlinear activation function, Loss function: in, is the i-th maintenance data, is the i-th predicted value of the ensemble model, λ KRR is the regularization parameter of kernel ridge regression, α i is the coefficient in kernel ridge regression, λ aRF is the regularization parameter of random forest, α j ′ and g j (x) are the coefficients and base model predictions of the jth tree in the random forest, is the variance term, γ is the weight parameter controlling the variance term, Check the contribution of input data to the ensemble model: Where v(S) represents the SHAP value of the subset S of the input data, and v(S∪{i}) represents the SHAP value of the new subset after adding the input data i. Remove input data with SHAP values less than the preset threshold, input the filtered input data into the integrated model, and use the predicted data of the integrated model as the influencing data; Based on the maintenance data, risk identification is performed on the deviation data to obtain a first weight of the influencing data, The maintenance data and deviation data are used to establish a data matrix. The maintenance data and deviation data in the data matrix correspond to each other one by one, and the vacant values are filled with 0. The deviation data is used to establish data points, and the data points are clustered: Neighborhood: N ∈,w(p) ={q∈D|dist(p,q)∈∧q≠p∧w(q)≥w min } Where D is the set of data points, w(q) is the local density weight of data point q, ∈ is the radius of the neighborhood, and w min is the weight threshold of the neighborhood, dist(p,q) represents the Euclidean distance between data point p and data point q, The core points of the neighborhood: Among them, N ∈,w(p) is the number of data points in the neighborhood, MinPts is the minimum point threshold used to determine whether a data point can become a core point, ν(p) is the noise term, α is the parameter for adjusting the noise term, and ν max is the maximum threshold of the noise term, is the local anomaly factor of the data point, Cluster Center Update: Among them, x is the data point in cluster j, C j is cluster j, I j is the influence term of cluster j, is the variance of the data points in cluster j, Objective function: Among them, β and γ are weight parameters, K is the number of clusters, λ is the weight of the regularization term, and F j is the compactness of cluster j, N noise is the distribution of noise points, R is the global regularization term, is the L1 norm, The failure rate in the cluster is calculated based on the maintenance data corresponding to the deviation data in the clustering result in the data matrix, which is used as the first weight. The product life cycle is normalized and preprocessed, and the average historical maintenance rate during the product life cycle is calculated through maintenance data based on the exponential decay model as the second weight affecting the data, and the average of the first weight and the second weight is calculated as the dynamic weight; A risk prediction model is established based on the impact data, the optimized dynamic weight and the maintenance data, the data of the smart product to be detected is input into the risk prediction model to output the risk prediction result, and a risk warning is performed.
2. According to the big data-based intelligent product quality and safety risk early warning method of claim 1, it is characterized in that: The performance data includes battery life, response time and functional indicators; user behavior data includes usage frequency, usage patterns and abnormal behaviors that cause smart product failures reported by users; product life cycle includes service life and update records; update records include the number of times the software and hardware of smart products have been updated; maintenance data includes the number of times the smart products have been repaired and the corresponding failure types; standard data is the average initial performance data of smart products.
3. According to the big data-based intelligent product quality and safety risk early warning method of claim 1, it is characterized in that: The method of comparing the performance data with the standard data to obtain deviation data comprises: Box plot analysis was performed using five numbers to summarize the data distribution. The box plot of performance data was compared with that of standard data. The box plot extended from the first quartile to the minimum non-outlier value, and the line from the third quartile to the maximum non-outlier value was the box plot whiskers. The box plot extended from the first quartile to the minimum non-outlier value, and the line from the third quartile to the maximum non-outlier value was the box plot whiskers. The performance data that exceeded the box plot whiskers were regarded as deviant data. The deviation value of the deviant data relative to the critical value of the standard data was calculated as the deviation data.
4. According to the big data-based intelligent product quality and safety risk early warning method of claim 1, it is characterized in that: The method for establishing a risk prediction model according to the impact data, the optimized dynamic weight and the maintenance data comprises: The weighted influence data is obtained by multiplying the influence data and the dynamic weight. The weighted influence data and the maintenance data are used to establish a data set. The data set is divided into a training set and a test set. The maintenance data is predicted based on the weighted influence data, and a regression model is established: Expected prediction results: Among them, Φ represents the probability distribution of the prediction results, is the predictor variable, are input features, represents the training set, is the set of model parameters of the regression model, Λ is the variational distribution of the model parameters of the regression model, Goodness of fit: Where Ξ represents the goodness of fit, ω is the parameter that adjusts the weight between the variational distribution and the likelihood function, ν is the model parameter, i is the index of ν, ρ is the variational distribution, Υ is the uniform prior distribution, Π is the likelihood function, ∈ is the uncertainty parameter, Probability density function: Among them, τ n is the noise term, is the predicted value of the regression model, α is the adjustment function, Goodness of fit for prediction uncertainty: Use the test set to test the accuracy of the regression model.
5. An intelligent product quality risk early warning system based on big data, used to execute the intelligent product quality risk early warning method based on big data according to any one of claims 1 to 4, characterized in that: include: The acquisition module acquires historical data of smart products, including performance data, user behavior data, product life cycle and maintenance data, and collects standard data; an optimization module, performing risk identification on the deviation data based on the maintenance data to obtain a first weight of the influencing data, and obtaining a dynamic weight based on the product life cycle and the first weight; a calculation module, comparing the performance data with the standard data to obtain deviation data, calculating the correlation degree of the user behavior data with the maintenance data to obtain influence data, and establishing a risk prediction model according to the influence data, the optimized dynamic weight and the maintenance data; The output module inputs the data of the smart product to be tested into the risk prediction model to output the risk prediction result and issue a risk warning.
6. An electronic device comprising: processor; as well as A memory arranged to store computer executable instructions, which when executed cause the processor to perform the method according to any one of claims 1 to 4.
7. A computer-readable storage medium storing one or more programs, which, when executed by an electronic device including a plurality of application programs, enables the electronic device to execute the method according to any one of claims 1 to 4.
Citation Information
Cited By
Risk early warning method and system based on data processing and machine learning
CN121352516A
