A method and system for analyzing risk characteristics of a domestic gas appliance throughout its life cycle

Abstract

The application relates to the field of gas appliance product quality safety, and particularly discloses a household gas appliance full-life cycle risk feature analysis method and system, which comprises the following steps: constructing a risk transmission topological structure, defining a multi-level root node state space and a conditional dependence relationship; based on the topological structure, performing step-by-step reasoning, obtaining product quality risk feature distribution in a production link, obtaining process coupling risk features in a test link, obtaining use link risk features in a use link, and obtaining accident risk features in an accident link; aggregating the use link risk features and the accident risk features, backtracking associated link risk features, constructing a cross-link risk feature correlation matrix, and outputting posterior risk feature parameters of each node in the full-life cycle. The application realizes step-by-step transmission and quantitative analysis of household gas appliance full-chain risk features from production to accidents, and provides quantitative basis for accurately positioning key nodes of risk transmission in each link.

Description

Technical Field

[0001] This invention relates to the field of gas appliance product quality and safety technology, specifically to a method and system for analyzing the risk characteristics of household gas appliances throughout their entire life cycle. Background Technology

[0002] In the field of risk management throughout the entire lifecycle of gas appliances, especially regarding the lifespan of household gas appliances and the transmission of risks across different stages, my country has long lacked unified and authoritative technical standards. Existing standards are outdated, resulting in the widespread use of products with potential risks, such as those exceeding their service life or having aging components, posing serious threats to residential gas safety. Furthermore, risk data from production, testing, use, and accidents is scattered across different responsible parties, lacking an effective mechanism for cross-stage risk transmission and aggregation. This leads to fragmented risk information, unclear transmission paths, and difficulty in accurately identifying the functions and responsibilities at each stage of the lifecycle, making it impossible to form a closed-loop management system covering design, production, on-site use, and accident feedback.

[0003] The patent publication number CN120181680A, entitled "A Method for Risk Assessment of Gas Appliance Product Quality," proposes risk assessment classification indicators for product quality, constructs an evaluation matrix, and determines the weight of individual evaluation indicators, achieving quality risk analysis from production to testing. However, this method only covers static evaluation at the production and testing stages, failing to consider the dynamic impact of the usage process on quality risk throughout the entire life cycle. Furthermore, it does not address the disturbance mechanisms of key factors such as uncontrolled usage behavior, nonlinear effects of product aging, and deviations in the accuracy of accident reports on risk characteristics. Existing technologies lack quantitative analysis methods for the complete transmission path of risk from quality formation, testing exposure, usage evolution to accident feedback, and cannot effectively reveal the correlation strength and sensitivity coefficients of risk characteristics at each stage, making it difficult to meet the needs of refined analysis of risk characteristics throughout the entire life cycle of household gas appliances.

[0004] Therefore, there is an urgent need for a risk characteristic analysis method that can cover the entire life cycle of household gas appliances. By constructing a cross-stage risk transmission topology, production, testing, use, and accident stages are incorporated into a unified assessment framework. The method establishes the conditional dependencies between the root nodes and risk characteristic nodes of each stage, quantifies the node associations and risk transmission paths, and realizes the step-by-step transmission and evolution analysis of risk characteristics from quality formation, risk exposure, injury occurrence to accident feedback. This provides a quantitative basis for accurately locating key nodes and responsibility boundaries of risk transmission in each stage, effectively overcoming the problems of fragmented industry standards and insufficient full-cycle risk governance. Summary of the Invention

[0005] The technical problem to be solved by the present invention is that the existing technology for risk analysis of household gas appliances is limited to static assessment of a single link and lacks a mechanism for the step-by-step transmission of risk characteristics and cross-link correlation quantification of the entire chain from production to accident. The technical solution of this invention to solve the above-mentioned technical problems is as follows: A method for analyzing the risk characteristics of the entire life cycle of household gas appliances, specifically including the following steps: S1. Define the root node state space covering the enterprise's survival time and reputation in the production stage, the testing strategy and normal use risk probability in the testing stage, the product usage status and demand quantity and service life in the usage stage, the total number of products and reporting accuracy in the accident stage, and the probability of serious injury and minor injury under uncontrolled use conditions and the probability of preventing injury in uncontrolled use conditions; establish the conditional dependencies between each root node and the corresponding risk characteristic nodes in the stage, forming a Bayesian network topology covering the production, testing, usage, and accident stages; S2. In the production stage, using the root node state space of the enterprise's duration of existence and reputation as input, the probability distribution of product quality at different levels is obtained through reasoning, and output as the product quality risk characteristic distribution. In the testing stage, using the root node state space of the product quality risk characteristic distribution, testing strategy, and normal use risk probability as input, the normal use risk probability is first corrected using the risk correction coefficients corresponding to different states of the testing strategy, and then further corrected using the process risk correction coefficients corresponding to different levels of product quality, to obtain the use risk probability distribution after dual correction by testing strategy and process factors, which is output as the process coupling risk characteristic. In the usage stage, the root node state space of the process coupling risk characteristic is used in conjunction with product usage, demand quantity within the product life cycle, and service life. Using space as input, the process coupling risk characteristics are first corrected with usage behavior correction coefficients corresponding to different states of product usage to obtain usage risk probabilities. Then, based on the demand quantity and usage risk probabilities within the product's life cycle, a probability model assuming at least one risk occurrence is used to calculate the risk probability within the life cycle. Next, a nonlinear aging risk correction model based on service life is introduced to correct the risk probability within the life cycle to obtain aging-corrected risk probabilities. Simultaneously, the probability of serious injury under uncontrolled conditions, the probability of minor injury under uncontrolled conditions, and the probability of preventing injury are used as root nodes to infer the probability of serious injury caused by risk and the probability of minor injury caused by risk. The aging-corrected risk probabilities are then fused with the probability of serious injury caused by risk and the probability of minor injury caused by risk to obtain the probability of major injury and the probability of minor injury, which are output as the risk characteristics of the usage stage. S3. In the accident stage, taking the root node state space of the probability of major injury and the probability of minor injury in the risk characteristics of the usage stage as input, the total number of products and the accuracy of the report, the total number of serious injuries and the total number of minor injuries are calculated based on the total number of products and the probability of major injury and the probability of minor injury. Then, the total number of serious injuries and the total number of minor injuries are corrected by the reporting deviation correction coefficient corresponding to different states of the reporting accuracy to obtain the actual number of serious injuries and the actual number of minor injuries. The probability of product causing injury, the probability of product causing serious injury, and the probability of product causing minor injury are calculated based on the ratio between the actual number of injuries and the total number of products, and are output as the accident risk characteristic distribution. The risk characteristics of the usage stage and the accident risk characteristic distribution are aggregated, and the product quality risk characteristic distribution and the process coupling risk characteristics are back-linked. The risk transmission path sensitivity coefficient and correlation strength between each risk characteristic are calculated, and a cross-stage risk characteristic correlation matrix is ​​constructed. The output is the posterior risk characteristic parameters covering each node of the entire life cycle of the production stage, testing stage, usage stage and accident stage.

[0006] A system for analyzing the full life cycle risk characteristics of household gas appliances, used to implement a method for analyzing the full life cycle risk characteristics of household gas appliances, including: The risk transmission topology construction module is configured to build a risk transmission topology structure for the entire life cycle of household gas appliances, defining the root node state space and conditional dependencies covering the production, testing, use, and accident stages. The step-by-step risk feature reasoning module is configured to reason step-by-step based on the topology, obtain the product quality risk feature distribution in the production stage, obtain the process coupling risk feature through double correction in the testing stage, obtain the aging correction risk probability through behavior correction, batch accumulation and nonlinear aging correction in the usage stage, and fuse the aging correction risk probability with the probability of serious injury and the probability of minor injury caused by risk under uncontrolled usage conditions to obtain the usage stage risk feature. The accident risk feature aggregation module is configured to calculate the total number of serious injuries and minor injuries at each stage of an accident and obtain accident risk features after deviation correction. It aggregates the risk features of the usage stage with the accident risk features and backtracks to associate the risk features of each stage, constructs a cross-stage risk feature association matrix, and outputs the posterior risk feature parameters for the entire life cycle.

[0007] The beneficial effects of this invention are: 1. This invention constructs a risk transmission topology covering the entire lifecycle of production, testing, use, and accidents, and establishes a conditional dependency relationship between multi-level root nodes and risk feature nodes. This enables step-by-step conditional probability reasoning of risk features from quality formation, testing exposure, use evolution to accident feedback, overcoming the shortcomings of existing technologies that only perform static assessments of a single stage and lack quantitative analysis of cross-stage risk transmission mechanisms.

[0008] 2. This invention incorporates the basic injury risk and protective capability under uncontrolled use conditions, the nonlinear effect of product aging, and the accuracy deviation of accident reports into a unified evolutionary framework. Through multiple coupled calculations of usage behavior correction, batch accumulation, aging correction, and reporting deviation correction, it solves the shortcomings of existing technologies in lacking in-depth analysis of niche technical issues such as uncontrolled use behavior, aging risk evolution, and statistical distortion.

[0009] 3. This invention constructs a cross-stage risk characteristic correlation matrix by aggregating the risk characteristic distribution of each stage, calculates and outputs the sensitivity coefficient and correlation strength quantification value of the risk transmission path, accurately locates the key nodes and responsibility boundaries of risk transmission in each stage, and provides a quantitative basis for closed-loop tracking, source analysis and control decision-making of risk characteristics throughout the entire life cycle of household gas appliances. Attached Figure Description

[0010] Figure 1 This is a schematic diagram of the steps in the whole life cycle risk characteristic analysis method described in this invention; Figure 2 This is the network diagram for calculating the probability of product quality nodes in this invention. Figure 3 The network diagram used for risk probability calculation when considering the process in this invention; Figure 4 This is a network diagram illustrating the risk characteristics analysis of the entire life cycle of household gas appliances according to the present invention. Detailed Implementation

[0011] The principles and features of the present invention are described below. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.

[0012] Example 1 like Figure 1-4 As shown, this invention provides a method for analyzing the risk characteristics of household gas appliances throughout their entire life cycle, specifically including the following steps: S1. Construct a risk transmission topology for the entire life cycle of household gas appliances, and define the root node state space and conditional dependencies covering the production, testing, use, and accident stages; S2. Based on the topology, reason step by step to obtain the product quality risk feature distribution in the production stage, and obtain the process coupling risk feature through double correction in the testing stage; obtain the aging correction risk probability through behavior correction, batch accumulation and nonlinear aging correction in the usage stage; and fuse the aging correction risk probability with the probability of serious injury and the probability of minor injury caused by risk under uncontrolled usage conditions to obtain the usage stage risk feature. S3. Calculate the total number of serious injuries and minor injuries at each stage of the accident, and obtain the accident risk characteristics after deviation correction; aggregate the risk characteristics of the usage stage and the accident risk characteristics, and backtrack to associate the risk characteristics of each stage to construct a cross-stage risk characteristic association matrix, and output the posterior risk characteristic parameters of the whole life cycle.

[0013] Example 2 Preferably, in Embodiment 1, step S1 is specifically implemented through the following steps: S11. Definition of the root node state space and setting of probability distribution in the production and testing phases are as follows: The enterprise survival duration node adopts a discrete distribution, set as 5 states, with the probability distribution satisfying: P(enterprise survival duration = k) = 1 / 5, k∈{<1 year, [1 year, 5 years), [5 years, 10 years), [10 years, 20 years], more than 20 years}. This node is used to characterize the operating time span of the manufacturing enterprise within the industry. The longer the operating time, the more mature the production quality management system usually is.

[0014] The enterprise credit node adopts a discrete distribution, set as 3 states, with a probability distribution satisfying: P(enterprise credit = k) = 1 / 3, k ∈ {no credit, good credit, relatively high credit}. Enterprise credit is classified into credit ratings according to GB / T23794, with Class A being relatively high credit, Class B being good credit, and Classes C and D being no credit. This node is used to reflect the potential impact of the enterprise's credit status on product quality.

[0015] The test strategy nodes are set using a discrete distribution to characterize the rigor of the methods used in product quality testing. The probability distribution is: P(test strategy k = below product testing standard) = 0.2, P(test strategy k = meets product testing standard) = 0.6, P(test strategy k = above product testing standard) = 0.2. As a factor influencing the adequacy of risk identification, the test strategy does not directly change the product's inherent risk level, but rather corrects the probability of risks during normal use by adjusting the degree of risk exposure.

[0016] Normal use of risk probability nodes It follows a truncated normal distribution with a cutoff interval of [0,1]. ,in This represents the average probability of a product experiencing a risk under normal usage conditions. The standard deviation represents the probability of risk; and Based on historical data such as the number of products released and the number of defective products under normal use, the Bayesian method was used to estimate and determine the quantity. The value ranges from 0.001 to 0.01, with a classic value of 0.005; The value ranges from 0.0005 to 0.002, with a classic value of 0.001.

[0017] S12. Use the state space definition and probability distribution setting of the root node of the process, specifically as follows: The product usage status nodes are set using a discrete distribution to characterize whether the product is used as expected. The probability distribution is as follows: P(product usage status k = expected usage) = 0.9, P(product usage status k = small deviation from expected usage) = 0.07, P(product usage status k = large deviation from expected usage) = 0.03. Small deviation from expected usage refers to a slight deviation from the design baseline in the usage scenario or operation method, causing local parameters such as temperature and stress to exceed the standard range but not reaching the safety limit, such as long-term use of oversized cookware or habitually sliding cookware forcefully instead of lifting it for placement. Large deviation from expected usage refers to usage methods that seriously violate physical principles or design limits, directly leading to irreversible physical damage or functional failure of key components, such as severe overflow of the pot without cleaning, clogging the ejector tube and nozzle.

[0018] Product lifecycle demand nodes This is used to characterize the total number of a specific type of product put into use throughout its entire life cycle and thus exposed to potential risks. The probability distribution follows a cutoff interval of [0, 10]. 8 The truncated normal distribution of ], i.e. ,in and Based on market statistics, The value ranges from 50 to 500, with a classic value of 100. The value ranges from 50 to 200, with a classic value of 100.

[0019] Useful years node The probability distribution used to characterize the actual service life of a product follows a truncated normal distribution with a cutoff interval of [0, 15]. ,in The location parameter of the normal distribution before truncation has a range of 3 to 8, with a classic value of 5; The scaling parameter is the untruncated normal distribution, ranging from 3 to 8, with a classic value of 5. This parameter setting ensures that the actual expected value of the truncated distribution is in the range of 4 to 6 years, which is consistent with the actual service life distribution of household gas appliances.

[0020] S13. Definition and probability distribution setting of the state space of the root node in uncontrolled use and accident scenarios, specifically as follows: The "Serious Injury Probability under Uncontrolled Use" node represents the fundamental probability uncertainty of serious injury occurring under uncontrolled use conditions. This node has 100 discrete states with a discrete uniform prior distribution, meaning the prior probability of each state is P(x) = 1 / 100, X5 ∈ [0,1], where X5 represents the probability of serious injury under uncontrolled use. The "Minor Injury Probability under Uncontrolled Use" node represents the fundamental probability uncertainty of minor injury occurring under uncontrolled use conditions. This node has 100 discrete states with a discrete uniform prior distribution, meaning the prior probability of each state is P(x) = 1 / 100, X6 ∈ [0,1], where X6 represents the probability of minor injury under uncontrolled use. The "Prevention of Injury Probability" node represents the fundamental probability uncertainty of preventing injury under uncontrolled use conditions. This node has 100 discrete states with a discrete uniform prior distribution, meaning the prior probability of each state is P(x) = 1 / 100, X7 ∈ [0,1], where X7 represents the probability of preventing injury. Total Product Node The uncertainty in the quantity of this type of product in use in the market is set to 10. 8 There are several discrete states, and their prior distribution is a discrete uniform distribution, meaning the prior probability of each state is P(x) = 1 / 10. 8 x∈{0,1,2,…,10 8 1}.

[0021] The report accuracy node is used to characterize the data deviation introduced by factors such as underreporting and false reporting during the accident statistics process. It adopts a discrete distribution setting, and its probability distribution is P(report accuracy k = lower than the number of reports) = 0.04, P(report accuracy k = the number of reports) = 0.95, P(report accuracy k = higher than the number of reports) = 0.01.

[0022] S14. Construction of the Bayesian network topology for risk propagation throughout the entire lifecycle, specifically as follows: Using the enterprise's survival duration and reputation nodes in the production process as parent nodes, and the product quality risk characteristic node as child nodes, a conditional dependency relationship is established for the production process. Using the testing strategy node and the normal usage risk probability node as parent nodes, and the normal usage risk probability node considering the testing strategy as child nodes, a primary conditional dependency relationship is established for the testing process. Then, using the product quality risk characteristic node and the normal usage risk probability node considering the testing strategy as parent nodes, and the process coupling risk characteristic node as child nodes, a secondary conditional dependency relationship is established for the testing process, achieving a two-layer correction of product quality on the risk characteristics of the testing process.

[0023] Using product usage status nodes, product lifecycle demand quantity nodes, and service life nodes as parent nodes, a chain-like conditional dependency relationship is established sequentially between usage risk probability nodes, lifecycle risk probability nodes, and aging correction risk probability nodes. Using uncontrolled serious injury probability nodes, uncontrolled minor injury probability nodes, and the probability of preventing injury as parent nodes, and risk leading to serious injury probability nodes and risk leading to minor injury probability nodes as child nodes, an uncontrolled usage conditional dependency relationship is established. The aging correction risk probability nodes are associated with risk leading to serious injury probability nodes and risk leading to minor injury probability nodes, respectively, with major injury probability nodes and minor injury probability nodes as fusion output child nodes, establishing a comprehensive conditional dependency relationship for the usage phase.

[0024] Using the total number of products and the report accuracy node as parent nodes, and the actual number of seriously injured and the actual number of slightly injured nodes as child nodes, a conditional dependency relationship for the accident process is established. The probability of serious injury and the probability of minor injury output from the usage stage are used as risk inputs for the accident process, with the total number of seriously injured and the total number of slightly injured nodes as intermediate child nodes, thus constructing a conditional dependency path for accident statistics and correction.

[0025] Through the aforementioned multi-level parent-child node associations, a Bayesian network topology covering the entire lifecycle of risk transmission across four stages: production, testing, use, and incidents is formed. The conditional dependencies within and between each stage are quantitatively characterized using conditional probability distributions or probability density functions. Specifically, the conditional probability distribution between the production and testing stages is determined by the scoring mapping and truncated normal distribution described in S21; the conditional probability distribution between the testing and use stages is determined by the weighted average of correction coefficients described in S22; the conditional probability distribution of uncontrolled use branches within the use stage is determined by the discrete state ergodicity described in S24; and the conditional probability distribution of the incident stage is determined by the reporting bias correction described in S32.

[0026] Example 3 Preferably, in Embodiment 1, step S2 is specifically implemented through the following steps: S21. Inference on the distribution of product quality risk characteristics in the production process, specifically including: S211. Map the five discrete states of the enterprise's duration of operation to enterprise duration scores. The mapping values ​​increase sequentially by years: 0.0, 0.25, 0.5, 0.75, and 1.0. Specifically, states less than 1 year are mapped to 0.0, states greater than or equal to 1 year but less than 5 years are mapped to 0.25, states greater than or equal to 5 years but less than 10 years are mapped to 0.5, states greater than or equal to 10 years but less than or equal to 20 years are mapped to 0.75, and states over 20 years are mapped to 1.0. The longer the duration, the higher the score, indicating that the maturity of the enterprise's quality management system increases with the length of operation.

[0027] The three discrete states of the enterprise credit node are mapped to enterprise credit scores, with 0.0 for no credit, 0.5 for good credit, and 1.0 for relatively high credit. Enterprise credit ratings are divided according to GB / T23794, with A being relatively high credit, B being good credit, and C and D being no credit.

[0028] The mean of the product quality distribution is calculated based on a weighted average of enterprise duration score and enterprise reputation score. The calculation formula is: In the formula: This indicates the score for the duration of a company's existence. This indicates the company's credit rating. Weighting based on the length of a company's existence. The weighting of corporate reputation is higher than that of corporate duration. Both weighting parameters are obtained using linear regression methods based on historical product quality and corporate data from the industry. Regression analysis based on historical product quality data of industry companies shows that the regression coefficient for corporate reputation is approximately 2.1 times that for duration, a classic value. For 1, The average value is 2, indicating that the predicted results match the actual product quality by 90%; this average value... It represents the expected level of product quality given the company's lifespan and reputation.

[0029] S212. The mean calculated using S211 To determine the expected value, the standard deviation is determined based on the degree of fluctuation in historical product quality data. Using the scaling parameter, a truncated normal distribution model is constructed for the product quality evaluation variables. The truncated normal distribution is defined within the value interval [0,1] of the product quality evaluation variables, i.e., the product quality evaluation variables... satisfy In the formula: The mean value calculated for S211. The standard deviation is a parameter used to ensure that the probability distribution of product quality is highly concentrated around the mean, reducing the impact of random fluctuations on the evaluation results. The value ranges from 0.0005 to 0.005, the classic value. Set to 0.001.

[0030] This truncated normal distribution model is used to characterize the continuous probability distribution of product quality under given enterprise survival duration and reputation. The lower limit of the truncation interval is 0, and the upper limit is 1, ensuring that the values ​​of product quality evaluation variables do not exceed a reasonable range, thereby guaranteeing the computational stability of the product quality risk characteristic distribution. The probability density function of the truncated normal distribution takes a value of 0 outside the truncation interval, and is normalized according to the normal distribution law within the truncation interval. Its probability density function expression is the normalized form of the cumulative distribution function difference at the boundary of the truncation interval.

[0031] S213. Based on product quality evaluation variables The value range of X1 is [0,1]. Product quality is divided into three levels: low, normal, and high, with boundaries of 0.333 and 0.666. That is, a low product quality level corresponds to a value of X1 ranging from 0 to 0.333, a normal product quality level corresponds to... The value ranges from greater than 0.333 to 0.666, corresponding to a relatively high product quality level. The value ranges from greater than 0.666 to 1. Based on the cumulative distribution function Φ of the truncated normal distribution, the probability of the product quality being at each level is calculated: The formula for calculating the probability that a product has a low quality grade is: The formula for calculating the probability that a product quality grade is normal is: The formula for calculating the probability of a product having a high quality grade is: The probability distribution of each level is used as the output of the product quality risk feature distribution, which is used as the basis for selecting the process risk correction coefficient in the subsequent process coupling risk feature calculation in the testing stage; this product quality risk feature distribution realizes the probability transfer from the root node state space of the production stage to the risk feature node.

[0032] S22. Dual correction reasoning for process coupling risk characteristics in the testing phase, specifically including: S221. Obtain the product quality risk characteristic distribution output from the production process. This distribution includes probability values ​​for product quality at three levels: low, normal, and high. These probabilities serve as input parameters for process risk correction. Call the state space of the root node of the test strategy defined in the topology. The root node of the test strategy includes three discrete states: below product test standards, meeting product test standards, and above product test standards. Correct the normal use risk probability using the risk correction coefficients corresponding to each state of the test strategy, obtaining the normal use risk probability considering the test strategy. The risk correction coefficients are set according to the test strategy states and satisfy... The corrected calculation formula is as follows: ,in: The probability of risk during normal use. To account for the probability of risk when using the testing strategy normally.

[0033] When considering the probability of risk during normal use of the testing strategy probability density satisfy In the formula: Testing strategy To test the probabilities of each state of the policy, Risk probability during normal use The probability density function, To account for the risk probability values ​​during normal use of the testing strategy, This represents the risk correction coefficient corresponding to the testing strategy state. The testing strategy, as a factor influencing the adequacy of risk identification, does not directly change the product's inherent risk level, but rather corrects the probability of risks during normal use by adjusting the degree of risk exposure.

[0034] S222. Obtain the risk probability of normal use when considering the test strategy output in S221, and the product quality risk characteristic distribution output from the production stage. Use the process risk correction coefficients corresponding to the three product quality levels to perform a secondary correction on the risk probability after the test strategy correction. The process risk correction coefficients are set according to the product quality level and satisfy... The corrected calculation formula is as follows: ,in: To account for the probability of risk during normal use of the testing strategy, To take into account the probability of risks used in the process.

[0035] When considering the probability of risk used in the process probability density satisfy , , In the formula: To consider the probability of risk used in the process The value of ; The product quality level status index is used to iterate through the three levels: low, normal, and relatively high. The test strategy state index iterates through three states: below, in line with, and above the product test standards. Product Quality For product quality to be at a certain level The probability of; Testing strategy The test strategy is in a state The probability of; Risk probability during normal use The probability density function; To test the policy state The corresponding risk adjustment factor; Product quality grade The corresponding process risk correction coefficient. This secondary correction reflects the impact of the production process control level on the risk during the usage stage. The process risk correction coefficient corresponding to a higher product quality level is smaller than that corresponding to a normal product quality level, and the process risk correction coefficient corresponding to a lower product quality level is larger than that corresponding to a normal product quality level.

[0036] S223. The probability distribution of usage risk after dual correction by testing strategy and product quality process factors is used as the output of process coupled risk feature for subsequent usage stage risk feature evolution calculation.

[0037] S23. Calculation of the probability evolution of aging correction risk in the usage process, specifically including: S231. Obtain the process coupling risk characteristics output from the testing phase. These characteristics represent the probability distribution of usage risks after dual correction by testing strategies and product quality process factors, and their probability density function is: The system invokes the root node state space of the product usage scenario defined in the topology. This root node state space, defined in S12, includes three discrete states: expected usage, small deviation from expected usage, and large deviation from expected usage. These correspond to three scenarios: normal product use according to design standards, slight deviation from design standards in usage scenarios or operating methods leading to local parameters exceeding standard ranges but not reaching safety limits, and usage methods severely violating design limits resulting in irreversible damage to critical components. The process coupling risk characteristics are corrected using usage behavior correction coefficients corresponding to each state to obtain the usage risk probability. Usage behavior correction coefficients. Based on the product's usage status, the settings should meet the requirements. The corrected calculation formula is as follows: ,in: As a characteristic of process coupling risks, To use risk probability.

[0038] Use risk probability probability density satisfy Product usage In the formula: (Product usage) ) represents the probability of each state of product usage. Characteristics of process coupling risks The probability density function, To use risk probability values, This is the usage behavior correction coefficient corresponding to the product usage status. This constraint ensures that the risk probability value after multiple corrections remains within the effective probability range. Product usage status is a key condition node reflecting the degree of deviation from actual usage behavior. When its correction coefficient is greater than 1, it indicates that the deviation in usage behavior leads to an increased degree of risk exposure; when it is equal to 1, it indicates that the product is used as expected. This achieves quantitative correction of the process coupling risk characteristics caused by usage behavior deviation.

[0039] S232. Based on the demand quantity and usage risk probability within the product lifecycle, a probability model assuming at least one risk occurrence is used to calculate the risk probability within the lifecycle. This model aligns with the actual risk propagation characteristics of mass use of household gas appliances and better reflects the actual risk accumulation pattern compared to simple linear superposition. Demand quantity within the product lifecycle. The total number of a specific type of product that is put into use and exposed to potential risks during its life cycle, as defined in S12, follows a cutoff interval of [0, 10]. 8 The product follows a truncated normal distribution. This is because the product lifecycle includes... Each product is independently exposed to risk, and the overall risk is the complement of the joint probabilities of each individual product being risk-free, calculated as follows: , in: To use risk probability, The quantity required over the product's life cycle is used in the calculation, and the expected value or a specific fixed value is substituted into it. This represents the probability of risk within the lifespan.

[0040] Risk probability during life cycle probability density Based on the probability of use risk probability density Derived using the variable transformation method, when the required quantity during the product life cycle is a fixed value. hour, and The functional relationship is ,in for The inverse function is given by the value of . Then the probability density transformation satisfies In the formula: To use risk probability The probability density function, The value is assigned to the probability of risk over the lifespan. The total number of products exposed to risk, i.e., the demand during the product's life cycle. The probability model uses the power operation of the risk-free probability of a single product to represent the joint risk-free probability, and uses the complement of the joint risk-free probability as the risk probability within the life cycle, accurately depicting the cumulative probability effect of at least one risk occurring when multiple products are independently exposed to risk.

[0041] S233. A nonlinear aging risk correction model based on service life is introduced to correct the risk probability within the lifespan, thus obtaining the aging correction risk probability. Service life The actual service life of the product, defined in S12, follows a truncated normal distribution with a cutoff interval of [0, 15]. Based on experimental data showing a quadratic nonlinear correlation between the aging risk of gas appliances and their service life, the failure probability of seals and gas pipelines increases exponentially with the increase in service life, calculated as follows: ,in: This represents the probability of risk over the lifespan. To determine the service life, an expected value or a specific fixed value is used in the calculation. To correct for the probability of aging-related risks. This is the product type index, with a value ranging from 1.8 to 2.2, and a classic value of 2.0.

[0042] Probability of aging correction risk probability density Risk probability within the life cycle and service life The joint distribution was derived using the variable transformation method, when the service life is taken as a fixed value. And when the correction result does not reach the upper limit truncation, and The functional relationship is Its inverse function is Then the probability density transformation satisfies In the formula: Risk probability within the lifespan The probability density function, The value is assigned to the probability of aging correction risk. The actual service life of the product, i.e., its service life. The value of , This is a product type index. When... hour, Take the upper limit value of 1.0, at which point... A probability mass point is generated at point O, and its probability mass is... In the formula: The actual service life of the product, i.e., its service life. The value of , Risk probability within the life cycle The probability density function represents the cumulative probability of risk reaching its upper limit after being amplified by aging within the lifespan. In this nonlinear aging risk correction model, the aging risk correction is the product of the risk probability within the lifespan and the term of the service life raised to the power of the term, and the aging correction risk probability is the upper limit cutoff value of the sum of the risk probability within the lifespan and the aging risk correction.

[0043] S24. Risk fusion of uncontrolled use conditions and generation of risk characteristics during use, specifically including: S241. Call the root node state spaces defined in the topology for the probability of serious injury under uncontrolled conditions, the probability of minor injury under uncontrolled conditions, and the probability of preventing injury. These three root node state spaces were set in S13. The probability nodes for serious injury and minor injury under uncontrolled conditions are each set to 100 discrete states with a prior uniform distribution. The probability node for preventing injury is also set to 100 discrete states with a prior uniform distribution. The values ​​of each node state are distributed in the range of 0 to 1. By calling the above three root node state spaces, the probability distribution parameters of basic injury risk and protection capability under uncontrolled use conditions are obtained, which serve as the prior input conditions for subsequent calculations of the probability of serious injury caused by risk and the probability of minor injury caused by risk.

[0044] S242. Using the probability nodes for severe injury under uncontrolled conditions and the probability node for preventing damage as parent nodes, and the probability node for risk leading to severe injury as child nodes, construct the conditional dependency relationship for severe injury occurrence under uncontrolled usage conditions. The probability node for severe injury under uncontrolled conditions is denoted as... The probability node that prevents damage is denoted as The node where the risk leads to serious injury is denoted as Based on the mechanism of serious injury under uncontrolled use, when the injury prevention mechanism fails, the risk of serious injury is realized. Therefore, the probability of serious injury resulting from this risk is the product of the probability of serious injury under uncontrolled use and the complement of the probability of injury prevention. The formula is as follows: ,because and Each state is set to 100 discrete states, and their state values ​​are as follows: and ,in ;but The possible values ​​are Assuming that the probability of serious injury and the probability of preventing injury are independent when the situation is uncontrolled, and that each discrete state follows a uniform distribution, then 100 state combinations form 10,000 equally probable basic events. The probability mass density of the risk leading to serious injury is: In the formula: and For discrete state indexing, for Possible values ​​for .

[0045] Similarly, the probability node for minor injury when out of control and the probability node to prevent damage As the parent node, the node with the probability of minor injury due to risk. For child nodes, the calculation formula is: The probability mass density of the risk leading to minor injury is: In the formula: The above calculations obtain the probability mass density through discrete state traversal, accurately characterizing the residual risk distribution of basic injury risk under uncontrolled use conditions after the protective capabilities have been weakened.

[0046] S243. Obtain the aging correction risk probability node after aging correction during the usage phase. This node represents the risk probability within the lifespan after considering aging factors, and its probability density function is: The aging-corrected risk probability is fused with the probability of serious injury and the probability of minor injury caused by the risk, respectively, to obtain the probabilities of major injury and minor injury. Major injury probability node. The product of the aging-related risk probability and the probability of serious injury resulting from the risk is used to determine the minor injury probability node. The product of the aging-corrected risk probability and the probability of the risk causing minor injury is calculated as follows: ,because and For discrete random variables, For continuous random variables, the probability of serious injury The probability density is derived from the law of total probability and the method of variable transformation.

[0047] for For each possible value of , the probability of serious injury given the risk is . At that time, the probability of serious injury The conditional probability density is ,in In the formula: Probability of serious injury The value of , Correcting the probability of aging The probability density function, The risk leads to a high probability of serious injury. The discrete possible values ​​are: and These are the discrete state indices for the nodes representing the probability of severe injury when the situation is out of control and the probability of preventing damage, respectively. The probability density of the probability of major injury is obtained by weighting and averaging all 10,000 discrete state combinations. .when When the probability of serious injury is 0, the corresponding risk is zero. It is also 0. When summing, this combination of states is directly assigned a probability mass of 0 and does not participate in the continuous probability density weighting.

[0048] Similarly, the probability of minor injury The probability density is In the formula: Probability of minor injury The value of is determined by combining the probability of major injury and the probability of minor injury as risk characteristics output during the usage phase, which are then used to calculate the expected value of the total number of injuries in subsequent accident phases.

[0049] Example 4 Preferably, in Embodiment 1, step S3 is implemented through the following steps: S31. Calculation of the expected total number of injuries in the accident process, specifically including: S311. Obtain the probability of significant damage output during the usage phase. and probability of minor injury Call the root node state space of the total number of products defined in the topology. The total number of products has been set using a discrete uniform distribution in S13, with the state space covering 0 to 10. 8 In the interval, the prior probability of each discrete state is 1 / 10. 8 The probability of major injury, the probability of minor injury, and the total number of products are used as prior input conditions for calculating the total number of injuries in the accident process.

[0050] S312. Model the number of people who suffer serious injuries and minor injuries during the product's use phase as random variables following a binomial distribution, with the total number of serious injuries being... The parameter is the total number of products. With the probability of major injury The binomial distribution, the total number of minor injuries The parameter is the total number of products. Probability of minor injury The binomial distribution.

[0051] The probability mass function of the binomial distribution is determined by the binomial distribution formula, namely: In the formula: This represents the total number of seriously injured individuals. The number of serious injuries. ; Total number of products; This represents the probability of causing significant damage. Let be the combination number, representing the number of combinations from . Selected from Taiwanese products The number of combinations that resulted in serious injuries in Taiwan. Given the total number of products. The magnitude is extremely large, and the standard deviation of the binomial distribution is negligible relative to the mean. Therefore, in calculating the proportion of accident links, the expected value is used as an approximation as the input value for the deterministic node. The calculation formulas are as follows: and In the formula For the total number of products, This represents the probability of causing significant damage. This represents the probability of minor injury. The expected value calculation achieves a deterministic transfer of risk characteristics from the usage stage to the injury quantity node at the accident stage.

[0052] S32. Accident Risk Characteristic Report Bias Correction and Distribution Output, specifically including: S321. Obtain the total number of critically injured characters output by S311 and S312. and total number of minor injuries The system calls the state space of the root node for reporting accuracy, which is already defined in the topology. The reporting accuracy node has been set in S13 to three discrete states: below the number of reports, the number of reports, and above the number of reports. The probabilities of each state are 0.04, 0.95, and 0.01, respectively, to characterize the data bias introduced by missed reports and false reports during the accident statistics process.

[0053] S322. Correct the total number of serious injuries and the total number of minor injuries using the reporting deviation correction factors corresponding to each state of reporting accuracy. Reporting deviation correction factor Based on the report accuracy status settings, meet the requirements. Actual number of seriously injured and the actual number of minor injuries The results are obtained by correcting for the total number of serious injuries and the total number of minor injuries using a reporting deviation correction factor, calculated as follows: , In the formula This represents the total number of seriously injured individuals. The total number of minor injuries, The reporting bias correction factor is used. When the correction factor is less than 1, it indicates that the number of injuries is underestimated due to underreporting, and when it is greater than 1, it indicates that the number of injuries is overestimated due to misreporting.

[0054] S323. Calculate the probability of injury caused by the product by dividing the sum of the actual number of seriously injured and the actual number of slightly injured by the total number of products; calculate the probability of serious injury caused by the product by dividing the actual number of seriously injured by the total number of products; calculate the probability of minor injury caused by the product by dividing the actual number of slightly injured by the total number of products. The probability of the product causing serious injury The probability of minor injury caused by the product. The calculation formulas are respectively , , In the formula: This represents the actual number of seriously injured. The actual number of people with minor injuries. This represents the total number of products.

[0055] S324. The probability of the product causing injury, the probability of the product causing serious injury, and the probability of the product causing minor injury are all output as accident risk characteristics. This accident risk characteristic represents the comprehensive injury risk probability distribution of the product at the on-site accident stage after correction for reporting bias. It forms an important component of the overall product risk assessment result and serves as the accident stage input parameter for constructing the cross-stage risk characteristic correlation matrix from S331 to S334.

[0056] S33. Construction of cross-stage risk feature association matrix and output of posterior parameters, specifically including: S331. Using the probability change of each level of the product quality risk characteristic distribution as the independent variable and the corresponding change of the process coupling risk characteristic as the dependent variable, calculate the risk transmission sensitivity coefficient from the production stage to the testing stage; the change is determined by the numerical perturbation method, that is, by perturbing the mean of the product quality distribution. Apply small perturbation Recalculate the change in probability for each level. , , And observe the corresponding change in the expected value of the process coupling risk characteristics. The ratio of the two values ​​is calculated as the sensitivity coefficient. For products with a low quality level, the sensitivity coefficient... satisfy ,in: This represents the probability change of a low product quality level. The expected change in the corresponding process coupling risk characteristic, and the perturbation step size. The value ranges from 0.001 to 0.01, with a classic value of 0.005.

[0057] Similarly, the sensitivity coefficients for product quality grades of normal and high are respectively... and This sensitivity coefficient characterizes the marginal impact of the distribution of product quality risk characteristics in the production stage on the process coupling risk characteristics in the testing stage. The larger the absolute value of the coefficient, the stronger the transmission effect of the probability fluctuation of this level on the risk characteristics of the downstream testing stage.

[0058] S332. Using the expected value of the process coupling risk characteristics as the independent variable and the expected value of the usage stage risk characteristics as the dependent variable, calculate the risk transmission sensitivity coefficient from the testing stage to the usage stage. The process coupling risk characteristics have already been output in S223, and their expected value is... The risk characteristics of the usage process have been output in S243, including the probability of serious injury x15 and the probability of minor injury x16, with expected values ​​of respectively. and By changing the expected value of the process coupling risk characteristics and observing the changes in the expected value of the risk characteristics in the usage stage, the ratio of the two is calculated as a sensitivity coefficient.

[0059] For the probability of serious injury, the sensitivity coefficient satisfy ,in This represents the expected change in the characteristics of process coupling risks. This represents the expected change in the probability of serious injury. For the probability of minor injury, the sensitivity coefficient... satisfy This sensitivity coefficient characterizes the marginal impact of the risk characteristics coupled with the testing process on the probability of major and minor injuries in the usage stage. It reflects the amplification or attenuation effect of the risk characteristics after dual correction by testing strategies and production process factors in the usage stage.

[0060] S333. Using the expected value of the risk characteristics at the usage stage as the independent variable and the expected value of the accident risk characteristics as the dependent variable, calculate the risk transmission sensitivity coefficient from the usage stage to the accident stage. The risk characteristics at the usage stage have already been output in S243, including the probability of serious injury. and probability of minor injury Their expected values ​​are respectively and The accident risk characteristics have been output in S324, including the probability of the product causing serious injury. The probability of minor injury caused by the product. Their expected values ​​are respectively and .

[0061] By changing the expected value of risk characteristics in the usage phase and observing the changes in the expected value of accident risk characteristics, the ratio between the two is calculated as a sensitivity coefficient. For the transmission of the probability of serious injury, the sensitivity coefficient... satisfy ,in This represents the expected change in the probability of serious injury. This represents the expected change in the probability of serious injury caused by the corresponding product. For the propagation of the probability of minor injury, the sensitivity coefficient... satisfy This sensitivity coefficient characterizes the marginal impact of risk characteristics in the usage phase on risk characteristics in the accident phase, reflecting the final transmission effect of risk characteristics after aging correction and uncontrolled use fusion, after correction for accident statistics and reporting bias.

[0062] S334. Combine the risk characteristics of the usage phase and the accident risk characteristics in the order of production, testing, use, and accident to form a full life cycle risk characteristic vector. Risk Characteristic Vector Includes key risk characteristic parameters for each stage, meeting the requirements. In the formula: The probability of each level of product quality risk characteristic distribution. The expected value of the process coupling risk characteristics. and This represents the expected values ​​of the probability of major injury and the probability of minor injury during use. , and The expected value of the probability of the product causing injury, serious injury, or minor injury during the accident.

[0063] Based on the full life cycle risk feature vector, the sensitivity coefficients calculated from S331 to S333 are correlated according to the transmission direction from production to testing, testing to use, and use to accident, thus constructing a cross-stage risk feature correlation matrix. The correlation matrix satisfies In the formula: Let be the vector of sensitivity coefficients between the product quality risk characteristic distribution and the process coupling risk characteristics. This represents the vector of sensitivity coefficients between process coupling risk characteristics and usage stage risk characteristics. This is a vector of sensitivity coefficients between the risk characteristics of the usage stage and the risk characteristics of the accident stage, with 0 representing a zero matrix. Through matrix operations, it outputs posterior risk characteristic parameters covering each node of the entire lifecycle—production, testing, usage, and accident stages—achieving a quantitative characterization and correlation strength analysis of cross-stage risk transmission paths.

[0064] Define the risk feature sub-vector of the production process Risk feature sub-vectors in the testing phase Using process risk feature sub-vectors Risk feature subvectors of accident process Then the posterior risk characteristic scalar of the testing phase. Composed of production process sub-vectors and sensitivity coefficient vectors The weighted sum is determined to satisfy Similarly, the posterior risk feature vector of the usage process is used. ; Post-accident risk feature vector The posterior risk characteristics of each stage are reorganized in sequence to form posterior risk characteristic parameter outputs covering all nodes of the entire life cycle.

[0065] Example 5 A system for analyzing the full life cycle risk characteristics of household gas appliances, used to implement a method for analyzing the full life cycle risk characteristics of household gas appliances, including: A. Risk transmission topology construction module, specifically including: The root node definition unit of the production process is configured to divide the enterprise's existence duration into multiple discrete states based on years, and the enterprise's reputation into multiple discrete states based on credit rating. The root node definition unit of the testing phase is configured to divide the testing strategy into three states according to the test standard level: below, in line with, and above the product testing standard, and configure the normal use risk probability to follow a truncated normal distribution. The root node of the usage process is defined as a unit that is configured to divide the product usage into three states based on the degree of deviation from the expected usage: expected usage, small deviation from expected usage, and large deviation from expected usage. The required quantity and service life of the product are configured to follow a truncated normal distribution. The uncontrolled root node definition unit is configured to set the state space by using discrete uniform distribution for the probability of serious injury when uncontrolled, the probability of minor injury when uncontrolled, and the probability of preventing injury. The root node definition unit of the accident process is configured to use a discrete uniform distribution for the total number of products and to divide the reporting accuracy into three states according to the degree of accident statistical deviation: below the number of reports, the number of reports, and above the number of reports. The topology generation unit is configured to use the root node defined in each stage as the parent node and the risk characteristics of the corresponding stage as the child nodes, establish the conditional dependencies between the nodes, and form a Bayesian network topology covering the production, testing, use, and accident stages.

[0066] B. The step-by-step risk characteristic reasoning module specifically includes: The production process reasoning unit is configured as a root node state space based on the enterprise's duration of existence and reputation. It calculates the mean of the product quality distribution through scoring mapping and weighted averaging. It constructs a truncated normal distribution model using the standard deviation determined based on the fluctuation of historical product quality data as the scale parameter. It divides product quality into three levels: low, normal, and high, calculates the probability of each level, and outputs the product quality risk characteristic distribution.

[0067] The inference unit in the testing phase is configured to obtain the product quality risk feature distribution output from the production phase as the input parameter for process risk correction. It calls the state space of the root node of the test strategy to correct the probability of normal use risk with the risk correction coefficient corresponding to each state, and then performs a second correction with the process risk correction coefficient corresponding to each level of product quality. The output is the process coupling risk feature after double correction by the test strategy and process factors.

[0068] Using the inference unit of the process stage, configured to obtain the process coupling risk characteristics output by the test stage, the root node state space of product usage is called to obtain the usage risk probability by correcting the usage behavior correction coefficients corresponding to each state. Based on the demand quantity within the product life cycle, the probability of risk within the life cycle is calculated using a probability model that assumes at least one risk occurrence. Then, a nonlinear aging risk correction model based on the service life is introduced for correction to obtain the aging correction risk probability.

[0069] The uncontrolled fusion unit is configured to call the root node state space of the uncontrolled serious injury probability, the uncontrolled minor injury probability, and the probability of preventing damage, which are defined in the topology. It calculates the risk leading to serious injury probability based on the product of the complement of the uncontrolled serious injury probability and the probability of preventing damage probability, and calculates the risk leading to minor injury probability based on the product of the complement of the uncontrolled minor injury probability and the probability of preventing damage probability. The aging correction risk probability is multiplied and fused with the above two probabilities respectively to obtain the major injury probability and the minor injury probability as the risk feature output of the usage stage.

[0070] C. Accident risk characteristic aggregation module, specifically including: The accident expectation calculation unit is configured to obtain the probability of serious injury and the probability of minor injury output in the usage stage, call the state space of the root node of the total number of products defined in the topology, calculate the total number of serious injuries based on the product relationship between the expected value of the total number of products and the probability of serious injury, and calculate the total number of minor injuries based on the product relationship between the expected value of the total number of products and the probability of minor injury.

[0071] The accident deviation correction unit is configured to obtain the total number of serious injuries and the total number of minor injuries. It calls the state space of the root node for reporting accuracy defined in the topology and corrects the total number of serious injuries and the total number of minor injuries with the reporting deviation correction coefficients corresponding to each state to obtain the actual number of serious injuries and the actual number of minor injuries. Then, it calculates the probability of injury caused by the product, the probability of serious injury caused by the product, and the probability of minor injury caused by the product based on the ratio of the actual number of injuries to the total number of products, and outputs the accident risk characteristics.

[0072] The correlation matrix construction unit is configured to use the ratios of the probability changes of each level of the product quality risk characteristic distribution to the corresponding changes of the process coupling risk characteristics, the ratios of the changes of the process coupling risk characteristics to the corresponding changes of the usage stage risk characteristics, and the ratios of the changes of the usage stage risk characteristics to the corresponding changes of the accident risk characteristics as sensitivity coefficients between each stage. The usage stage risk characteristics and the accident risk characteristics are concatenated in the order of the stages to form a full life cycle risk characteristic vector. Based on this vector, the sensitivity coefficients are correlated according to the transmission direction of the production, testing, usage, and accident stages to construct a cross-stage risk characteristic correlation matrix and output posterior risk characteristic parameters covering all nodes of the full life cycle.

[0073] Example 6 The specific implementation process of the above-mentioned technical solution of the present invention will be further described below with reference to a specific embodiment: Define the specific state values ​​and probability distribution parameters of each root node in the risk transmission topology of a certain model of household gas stove throughout its entire life cycle. For the production stage, the root node state space is set as follows: the enterprise's duration is less than 1 year, corresponding to an enterprise duration score. The enterprise's credit status is set to "Good," corresponding to the enterprise credit score. Weighting of Enterprise Duration Corporate reputation weight The root node state space setting for the testing phase: The test strategy state conforms to the product testing standards, with a corresponding risk correction coefficient. The average value of the risk probability parameter during normal use is taken. Standard deviation The root node state space setting for the usage phase: The product usage state is taken as the expected usage, with a corresponding usage behavior correction coefficient. To demonstrate the cumulative risk effect of mass production, the required quantity over the product lifecycle is taken as a fixed value. In practical applications, the demand quantity during the product's life cycle can be randomly sampled from a truncated normal distribution based on market statistics or taken as the expected value; the service life is taken as 3 years; and the probability of serious injury under uncontrolled conditions is taken as... When out of control, the probability of minor injury is taken. The probability of preventing damage is taken The root node state space of the accident process is set as follows: the total number of products is a fixed value. The accuracy status of the report is determined by the number of reports submitted, and the corresponding report deviation correction factor. The mean of the product quality distribution is calculated based on a weighted average of enterprise duration score and enterprise reputation score. , with mean For expectation and standard deviation Construct a truncated normal distribution model for the product quality evaluation variables based on the scaling parameters: Due to standard deviation The cumulative distribution function is extremely small and highly concentrated around the mean of 0.333, based on a truncated normal distribution. Calculate the probability of each product quality level. The probability of a product having a low quality level is... The probability that the product quality grade is normal is: The probability of a product having a high quality level approaches 0. The probability distribution of each level is used as the output of the product quality risk characteristic distribution, i.e., the probability of a low level. Normal level probability Higher probability .

[0074] The probability of normal use risk is adjusted using the risk correction coefficient corresponding to the test strategy status. The test strategy status is compliant with product testing standards, corresponding to... The probability of risk during normal use is taken as the expected value. Then, considering the risk probability of normal use when using the testing strategy, it is... A secondary correction is performed using the process risk correction coefficient corresponding to each product quality level. When the product quality level is low, the corresponding... Normally corresponds to When it is higher, it corresponds to Based on the low-to-medium probability of product quality risk characteristics distribution. and normal level probability The weighted average is applied, and the process risk correction coefficient is the weighted average value. Then, considering the risk probability used in the process, it is... The double-corrected probability distribution of usage risk is output as the process coupling risk feature, and its expected value is... The process coupling risk characteristics are corrected using a usage behavior correction coefficient corresponding to the product usage status. The product usage status is the expected usage, corresponding to... Then the risk probability is Based on the quantity required during the product lifecycle. and the probability of using risk The probability of risk over the lifespan is calculated using a probability model that indicates at least one risk occurrence. .because Therefore A nonlinear aging risk correction model based on service life is introduced, with a service life of 3 years and a product type index. The aging risk correction is then... The probability of aging correction risk is .

[0075] The probability of serious injury when out of control and the probability of preventing damage Calculate the probability of serious injury caused by the risk to the parent node: The probability of minor injury when out of control. and the probability of preventing damage Calculate the probability of minor injury caused by the risk to the parent node: .

[0076] By fusing the aging-corrected risk probability with the probability of serious injury and the probability of minor injury caused by the risk, the probabilities of major injury and minor injury are obtained: , The probability of major injury and the probability of minor injury are used as risk characteristics output during use. This is based on the total number of products. and probability of major damage Calculate the total number of seriously injured: Based on the total number of products and probability of minor injury Calculate the total number of minor injuries: .

[0077] The total number of serious injuries and the total number of minor injuries are corrected using the reporting deviation correction factor corresponding to the reporting accuracy status. The reporting accuracy status is the number of reports, corresponding to... The actual number of seriously injured is The actual number of people with minor injuries was: The probability of injury caused by the product is calculated by dividing the sum of the actual number of seriously injured and the actual number of slightly injured by the total number of products. The probability of a product causing serious injury is calculated by dividing the actual number of seriously injured persons by the total number of products. The probability of a product causing minor injuries is calculated by dividing the actual number of people with minor injuries by the total number of products sold. The probability of product-induced injury, the probability of product-induced serious injury, and the probability of product-induced minor injury are collectively used as the output accident risk characteristics. Risk characteristics from the usage stage and accident risk characteristics are aggregated and retrospectively correlated across different stages to construct a cross-stage risk characteristic correlation matrix. The change in low-level probabilities in the product quality risk characteristic distribution is used as the output. As the independent variable, the change in the corresponding process coupling risk characteristics. Calculate the sensitivity coefficient for the dependent variable: The variation in risk characteristics coupled with process The independent variable is the change in the probability of serious injury. Calculate the sensitivity coefficient for the dependent variable: The change in the probability of major injury during use. The independent variable is the change in the probability of serious injury caused by the corresponding product. Calculate the sensitivity coefficient for the dependent variable: The risk characteristics of the usage stage and the accident risk characteristics are concatenated in the order of the stages to form a full life cycle risk characteristic vector. The sensitivity coefficients are then correlated according to the transmission direction of the production, testing, usage, and accident stages to construct a cross-stage risk characteristic correlation matrix. The output is the posterior risk characteristic parameters of each node in the full life cycle covering the production, testing, usage, and accident stages.

[0078] The embodiments and / or implementation methods described above are merely preferred embodiments and / or implementation methods for implementing the technology of the present invention, and are not intended to limit the implementation methods of the technology of the present invention in any way. Any person skilled in the art can make some modifications or alterations to other equivalent embodiments without departing from the scope of the technical means disclosed in the present invention, but these should still be regarded as the technology or embodiments that are substantially the same as the present invention. This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. The above descriptions are only preferred embodiments of this application. It should be noted that due to the limitations of written expression, while there are objectively infinite specific structures, those skilled in the art can make several improvements, modifications, or changes without departing from the principles of this application, and can also combine the above technical features in an appropriate manner. These improvements, modifications, changes, or combinations, or the direct application of the inventive concept and technical solution to other situations without modification, should all be considered within the scope of protection of this application.

Claims

1. A method for analyzing the risk characteristics of a household gas appliance throughout its entire life cycle, characterized in that, include: Construct a risk transmission topology structure for the entire life cycle of household gas appliances, and define the root node state space and conditional dependencies covering the production, testing, use, and accident stages; Based on the aforementioned topology, the product quality risk characteristic distribution is obtained in the production stage, and the process coupling risk characteristics are obtained through double correction in the testing stage. The probability of aging correction risk is obtained through behavioral correction, batch accumulation, and non-linear aging correction during the usage phase. The aging correction risk probability is fused with the probability of serious injury and the probability of minor injury caused by risk under uncontrolled use conditions to obtain the risk characteristics of the use phase. The total number of serious injuries and minor injuries is calculated at each stage of the accident, and the accident risk characteristics are obtained after deviation correction. By aggregating the risk characteristics of the usage stage and the accident risk characteristics, and retrospectively associating the risk characteristics of each stage, a cross-stage risk characteristic correlation matrix is ​​constructed, and the posterior risk characteristic parameters of the entire life cycle are output.

2. The method for analyzing the full life-cycle risk characteristics of household gas appliances according to claim 1, characterized in that, The aforementioned construction of a risk transmission topology for the entire lifecycle of household gas appliances defines the root node state space and conditional dependencies covering production, testing, use, and accident stages, specifically including: The duration of a company's existence is divided into multiple discrete states based on years, and the company's reputation is divided into multiple discrete states based on credit rating. The testing strategy is divided into three states based on the level of testing standards: below, compliant, and above the product testing standards. The probability of risk during normal use follows a truncated normal distribution. Product usage is categorized into three states based on the degree of deviation from expected usage: expected usage, small deviation from expected usage, and large deviation from expected usage. The demand quantity and service life within the product's life cycle both follow a truncated normal distribution. The probability of serious injury when out of control, the probability of minor injury when out of control, and the probability of preventing injury are all set in a state space using a discrete uniform distribution. The total number of products is set in a discrete uniform distribution. The accuracy of the report is divided into three states according to the degree of deviation in accident statistics: below the number of reports, the number of reports, and above the number of reports. Using the root node defined in each stage as the parent node and the risk characteristics of the corresponding stage as the child nodes, conditional dependencies between the nodes are established to form a Bayesian network topology covering the production, testing, use, and accident stages.

3. The method for analyzing the full life-cycle risk characteristics of household gas appliances according to claim 1, characterized in that, Based on the aforementioned topological structure, a hierarchical reasoning process is used to obtain the product quality risk characteristic distribution in the production process, specifically including: Multiple discrete states of enterprise duration are mapped to enterprise duration scores, and multiple discrete states of enterprise reputation are mapped to enterprise reputation scores. The mean of the product quality distribution is calculated based on the weighted average of the enterprise duration scores and enterprise reputation scores. Using the mean as the expected value and the standard deviation determined based on the fluctuation of historical product quality data as the scale parameter, a truncated normal distribution model of the product quality evaluation variable is constructed. The truncated normal distribution is defined within the value range [0,1] of the product quality evaluation variable. Based on the value range of the product quality evaluation variables, the product quality is divided into three levels: low, normal, and high. The probability of the product quality being at each level is calculated based on the cumulative distribution function of the truncated normal distribution, and the probability distribution of each level is used as the output of the product quality risk characteristic distribution.

4. The method for analyzing the full life-cycle risk characteristics of household gas appliances according to claim 1, characterized in that, The process coupling risk characteristics obtained through double correction during the testing phase specifically include: The product quality risk characteristic distribution output from the production process is obtained as the input parameter for process risk correction. The state space of the root node of the test strategy defined in the topology is called to correct the probability of normal use risk with the risk correction coefficient corresponding to each state. The risk probability after the testing strategy correction is further corrected by the process risk correction coefficient corresponding to the three levels of product quality. The probability distribution of usage risk, after being modified by both testing strategies and product quality process factors, is output as the process-coupled risk feature.

5. The method for analyzing the full life-cycle risk characteristics of household gas appliances according to claim 1, characterized in that, The aging correction risk characteristics obtained during the usage phase through behavior correction, batch accumulation, and non-linear aging correction specifically include: The process coupling risk characteristics output from the testing phase are obtained. The product usage root node state space defined in the topology is called, and the process coupling risk characteristics are corrected with the usage behavior correction coefficients corresponding to each state to obtain the usage risk probability. Based on the demand quantity and usage risk probability within the product life cycle, the probability of risk within the life cycle is calculated using a probability model that assumes at least one risk. The probability model that assumes at least one risk uses the power operation of the probability of no risk for a single product to represent the joint probability of no risk, and uses the complement of the joint probability of no risk as the probability of risk within the life cycle. A nonlinear aging risk correction model based on service life is introduced to correct the risk probability within the service life period, thereby obtaining the aging correction risk probability. In the nonlinear aging risk correction model, the aging risk correction amount is the product of the risk probability within the service life period and the power of the service life period, and the aging correction risk probability is the upper limit cutoff value of the sum of the risk probability within the service life period and the aging risk correction amount.

6. The method for analyzing the full life cycle risk characteristics of household gas appliances according to claim 1, characterized in that, The aging correction risk probability is fused with the probability of serious injury and the probability of minor injury caused by risk under uncontrolled use conditions to obtain the risk characteristics of the use phase, specifically including: Call the root node state space of the uncontrolled serious injury probability, the uncontrolled minor injury probability, and the probability of preventing injury that are already defined in the topology; Using the probability of serious injury and the probability of preventing injury when out of control as parent nodes, the probability of serious injury caused by risk is calculated based on the product of the complements of the probability of serious injury when out of control and the probability of preventing injury; using the probability of minor injury and the probability of preventing injury when out of control as parent nodes, the probability of minor injury caused by risk is calculated based on the product of the complements of the probability of minor injury when out of control and the probability of preventing injury. The aging correction risk probability is multiplied by the probability of serious injury caused by the risk and the probability of minor injury caused by the risk, respectively, to obtain the probability of major injury and the probability of minor injury, which are output as the risk characteristics of the usage stage.

7. The method for analyzing the full life-cycle risk characteristics of household gas appliances according to claim 1, characterized in that, The calculation of the total number of serious injuries and the total number of minor injuries at the accident stage specifically includes: Obtain the probability of major injury and the probability of minor injury output during the usage phase, as well as the root node state space of the total number of products defined in the topology; The total number of serious injuries is calculated based on the product of the expected total number of products and the probability of major injury; the total number of minor injuries is calculated based on the product of the expected total number of products and the probability of minor injury.

8. The method for analyzing the full life cycle risk characteristics of household gas appliances according to claim 7, characterized in that, The accident risk characteristics obtained after deviation correction specifically include: Obtain the total number of serious injuries and the total number of minor injuries, as well as the state space of the root node for reporting accuracy defined in the topology; The total number of serious injuries and the total number of minor injuries are corrected by the reporting deviation correction coefficient corresponding to each state of the accuracy of the report, so as to obtain the actual number of serious injuries and the actual number of minor injuries. The probability of injury caused by the product is calculated by dividing the sum of the actual number of seriously injured and the actual number of slightly injured by the total number of products. The probability of serious injury caused by the product is calculated by dividing the actual number of seriously injured by the total number of products. The probability of minor injury caused by the product is calculated by dividing the actual number of slightly injured by the total number of products. The probability of the product causing injury, the probability of the product causing serious injury, and the probability of the product causing minor injury are used together as the accident risk characteristics output.

9. The method for analyzing the full life cycle risk characteristics of household gas appliances according to claim 8, characterized in that, By aggregating the risk characteristics of the usage phase and the accident risk characteristics, and retrospectively correlating the risk characteristics of each phase, a cross-phase risk characteristic correlation matrix is ​​constructed, and the posterior risk characteristic parameters for the entire life cycle are output, specifically including: Using the probability change of each level state in the product quality risk characteristic distribution as the independent variable and the corresponding change of the process coupling risk characteristic as the dependent variable, the ratio of the two is calculated as the sensitivity coefficient of the product quality risk characteristic distribution to the process coupling risk characteristic. Using the change in the process coupling risk characteristics as the independent variable and the corresponding change in the usage stage risk characteristics as the dependent variable, the ratio of the two is calculated as the sensitivity coefficient of the process coupling risk characteristics to the usage stage risk characteristics. Using the change in the risk characteristics of the usage phase as the independent variable and the corresponding change in the accident risk characteristics as the dependent variable, the ratio of the two is calculated as the sensitivity coefficient of the risk characteristics of the usage phase to the accident risk characteristics. The risk characteristics of the usage stage and the accident risk characteristics are concatenated in the order of the stages to form a full life cycle risk characteristic vector. Based on the full life cycle risk characteristic vector, the sensitivity coefficients are correlated according to the transmission direction of the production, testing, usage and accident stages to construct a cross-stage risk characteristic correlation matrix and output the posterior risk characteristic parameters covering all nodes of the full life cycle.

10. A system for analyzing the full life-cycle risk characteristics of household gas appliances, characterized in that, A method for analyzing the full life-cycle risk characteristics of household gas appliances as described in any one of claims 1-9 includes: The risk transmission topology construction module is configured to build a risk transmission topology structure for the entire life cycle of household gas appliances, defining the root node state space and conditional dependencies covering the production, testing, use, and accident stages. The step-by-step risk feature reasoning module is configured to reason step-by-step based on the topology, obtain the product quality risk feature distribution in the production stage, obtain the process coupling risk feature through double correction in the testing stage, obtain the aging correction risk probability through behavior correction, batch accumulation and nonlinear aging correction in the usage stage, and fuse the aging correction risk probability with the probability of serious injury and the probability of minor injury caused by risk under uncontrolled usage conditions to obtain the usage stage risk feature. The accident risk feature aggregation module is configured to calculate the total number of serious injuries and minor injuries at each stage of an accident and obtain accident risk features after deviation correction. It aggregates the risk features of the usage stage with the accident risk features and backtracks to associate the risk features of each stage, constructs a cross-stage risk feature association matrix, and outputs the posterior risk feature parameters for the entire life cycle.

Images