Interactive manual fault diagnosis system based on artificial intelligence

Abstract

The invention relates to the technical field of fault diagnosis, in particular to an artificial intelligence-based interactive manual fault diagnosis system, which comprises a fault state intelligent coding module, an intelligent fault correlation analysis module, a fault prediction intelligent analysis module and an intelligent fault response strategy module. According to the invention, automatic standardization of fault data is realized through intelligent classification and coding and an equipment driving assembly, the consistency and comparability of the data are improved, the data arrangement time is shortened in the automatic process, the response speed of fault diagnosis is accelerated, and intelligent analysis of a control circuit helps to construct a conditional fault probability matrix. Fault prediction accuracy is enhanced, fault identification is strengthened by combining fault prediction analysis of a vibration sensor and real-time operation data, a maintenance team can take preventive measures before faults occur, the downtime of equipment is remarkably shortened, and the adaptability of the maintenance measures is improved through real-time monitoring and strategy adjustment of an intelligent fault response strategy. The equipment operation efficiency is ensured; the service life is prolonged.

Description

Technical Field

[0001] The present invention relates to the technical field of fault diagnosis, and in particular to an interactive manual fault diagnosis system based on artificial intelligence. Background Art

[0002] The field of fault diagnosis technology focuses on identifying and determining problems and sources of failure in systems, equipment or software. The technology covers a variety of methods from simple fault detection to complex predictive maintenance strategies. In modern industrial and computing technology, fault diagnosis not only relies on traditional monitoring equipment, but also widely adopts advanced data analysis technologies such as machine learning and pattern recognition to improve the accuracy and efficiency of diagnosis. By analyzing the data of system operation, fault diagnosis technology can predict potential failures and warn in advance, thereby helping maintenance teams take timely repair measures, reduce downtime, and optimize resource utilization.

[0003] Among them, the interactive manual fault diagnosis system is a tool that uses an interactive user manual to help users identify and solve product or system failures. This system integrates a user interface, allowing users to input symptoms and related information of the problem through simple interactions. The intelligent algorithm behind the system will analyze the input, refer to the failure cases and solutions in the database, and provide users with specific diagnosis results and repair suggestions. The main purpose of this type of system is to provide immediate fault diagnosis support, reduce dependence on professional technicians, and improve maintenance efficiency and equipment availability.

[0004] Existing technologies improve the accuracy and efficiency of diagnosis through methods such as machine learning and pattern recognition, but the technology still relies heavily on a large amount of sample data and historical fault cases collected in advance, which limits its diagnostic capabilities when facing unknown or rare fault types. Existing fault diagnosis technologies mostly rely on traditional monitoring equipment, which is not sensitive enough to capture small or early abnormal behaviors, resulting in untimely identification of faults, which in turn affects the formulation and implementation of overall maintenance strategies. For example, a slight vibration anomaly has caused damage to key mechanical components before it is detected and diagnosed in time. This shortcoming is particularly prominent in modern fast-running production lines and high-demand industrial applications. Even a short downtime can lead to significant production losses and safety risks. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose an interactive manual fault diagnosis system based on artificial intelligence.

[0006] In order to achieve the above object, the present invention adopts the following technical scheme: An interactive manual fault diagnosis system based on artificial intelligence comprises:

[0007] The fault status intelligent coding module intelligently classifies and codes equipment faults based on the device driver component, obtains fault type identification, and converts historical fault data into standardized codes to form a standardized fault data set;

[0008] The intelligent fault correlation analysis module uses the standardized fault data set to intelligently analyze the correlation rules between fault types through the control circuit, generates a fault correlation diagram, calculates the conditional probability of fault occurrence based on the fault correlation diagram, and generates a conditional fault probability matrix;

[0009] The fault prediction intelligent analysis module applies the conditional fault probability matrix, based on the vibration sensor data, combined with the real-time operation data of the current equipment, intelligently predicts the occurrence of potential faults, generates predicted fault types, evaluates the impact and urgency of the faults based on the predicted fault types, and generates fault impact assessment results;

[0010] The intelligent fault response strategy module designs targeted intelligent fault response measures according to the fault impact assessment results, generates a fault response plan, implements the fault response plan and monitors the effects, adjusts measures and matches the real-time situation, and generates fault handling feedback results.

[0011] As a further solution of the present invention, the step of acquiring the standardized fault data set is specifically as follows:

[0012] Based on the device driver component, the device faults are intelligently classified and coded, the fault codes are located through the index table, and the status information is updated using the formula:

[0013]

[0014] Calculate and generate the weighted fault type index, where H ai represents the identifier of the i-th fault type, W ai represents the weight parameter of the i-th fault type, n a Represents the total number of fault types, T a represents the weighted fault type index;

[0015] The weighted fault type index is calculated with the historical fault data using the formula:

[0016]

[0017] The standard deviation index is calculated, where H ai Represents the identifier of the i-th fault type, T a represents the weighted fault type index, S a Represents the standard deviation indicator, n a Represents the number of fault types;

[0018] Each fault identification is normalized using the standard deviation index, using the formula:

[0019]

[0020] Each fault data is standardized to generate a standardized fault data set, where H ai Represents the identifier of the i-th fault type, T a represents the weighted fault type index, S a represents the standard deviation indicator, D a Represents the normalized fault dataset.

[0021] As a further solution of the present invention, the step of obtaining the fault association graph is specifically as follows:

[0022] The standardized fault data set is used for preliminary analysis, and the fault type and frequency are processed. The formula is:

[0023]

[0024] Generate fault impact index, where H bi represents the standardized data of the i-th fault type, μ b is the average value of the fault data, σ b is the standard deviation, I b Indicator indicating the degree of impact of the fault;

[0025] The fault impact index is applied to the intelligent control circuit, the fault correlation potential is calculated, and the nonlinear function is used to process the data formula:

[0026]

[0027] Generate a fault correlation potential index, where L b represents the fault correlation potential index, I b Represents the fault impact index, k b and t b Indicates adjustment parameters;

[0028] According to the fault correlation potential index, the correlation strength between the differentiated fault types is calculated by statistical analysis method, using the formula:

[0029] G b =∑L b ·E bi

[0030] Generate a fault correlation graph, where G b represents the fault correlation diagram, L b represents the fault correlation potential index, E biIndicates the existence of the connection between the i-th fault type.

[0031] As a further solution of the present invention, the step of obtaining the conditional failure probability matrix is ​​specifically as follows:

[0032] The weights of the interconnections between nodes of multiple fault types are extracted from the fault association graph, and the weighted sum of the edges corresponding to each fault node is calculated using the formula:

[0033]

[0034] Generate the weighted sum of the faulty nodes, where E bi represents the existence of the connection between the i-th fault type, P cj represents the probability weight of the connection, d c is the attenuation factor used to adjust the weight influence, W c represents the weighted sum of the faulty nodes;

[0035] According to the weighted sum of the fault nodes, the weighted normalization formula is applied to calculate the conditional probability of the fault occurring, through the formula:

[0036]

[0037] Generate the conditional probability of failure occurrence, where P fc represents the conditional probability of failure, W c represents the weighted sum of the faulty nodes, α c represents the normalization coefficient;

[0038] Using the conditional probability of the occurrence of the fault, the dependency between the differentiated faults is identified, the prerequisite for each fault to be triggered is determined, and the conditional probability values ​​between multiple faults are calculated, through the formula:

[0039] M fc =[P fc1 ,P fc2 ,…,P fcn ]

[0040] Construct the conditional failure probability matrix, where P fcn represents the conditional probability of the target fault type, P fc1 , P fc2 represents the conditional probability of the first and second fault types, M fc Represents the conditional failure probability matrix.

[0041] As a further solution of the present invention, the step of obtaining the predicted fault type is specifically as follows:

[0042] The probability vector associated with the current equipment type is extracted from the conditional failure probability matrix and adjusted for the impact of real-time operational data using the formula:

[0043]

[0044] Generate an adjusted failure probability vector, where M fc represents the conditional failure probability matrix, O dk Represents the influence coefficient of the operating data, c d is a constant used to adjust the effect, r d is the enhancement factor, V d represents the adjusted failure probability vector;

[0045] Based on the adjusted fault probability vector and combined with the vibration sensor data, the formula is:

[0046]

[0047] Generate a failure index where β d and γ d is the adjustment coefficient, p d is the nonlinear weighting index of vibration data, δ d is the normalization coefficient, V d represents the adjusted failure probability vector, S d Represents vibration sensor data, I d represents the fault index;

[0048] Using the fault index, the threshold determination method is applied to determine the fault type, through the formula:

[0049] T d =max(I d )>θ d

[0050] If T d Exceeding a predetermined threshold value θ d , then it is identified as the corresponding fault type and the predicted fault type is generated, where I d represents the fault index, T d is the fault type determination threshold, θ d Indicates a predetermined threshold.

[0051] As a further solution of the present invention, the step of obtaining the fault impact assessment result is specifically:

[0052] From the predicted fault types, associated fault impact parameters, including production downtime and quality loss, are extracted using the formula:

[0053]

[0054] Generate a fault impact score, where F hm is the severity of the fault impact, G hm is the loss caused by the failure, d h is the adjustment factor for the fault frequency, t h is the nonlinear adjustment coefficient affecting the depth, P h represents the fault impact score;

[0055] The fault impact score is combined with the current equipment maintenance and operation status data, through the formula:

[0056]

[0057] The fault impact index is obtained, where κ h and μ h are the coefficients for extracting weights from the equipment status and fault impact score, respectively. h Represents the health status of the device, h is the normalization factor, P h is the fault impact score, I h represents the fault impact score;

[0058] According to the fault impact index, a logical judgment method is used to determine the fault impact level and the priority of emergency treatment. h is greater than the set threshold θ h , the result is "high impact", otherwise it is "manageable impact", and the fault impact assessment result is generated.

[0059] As a further solution of the present invention, the steps of obtaining the fault response plan are specifically as follows:

[0060] Analyze the fault impact assessment results to determine the urgency and priority of the fault response using the formula:

[0061]

[0062] Get the fault response urgency, where E g Indicates the fault impact level, γ g is a coefficient adjusted based on device category and impact level, ∈ g is a fixed value adjusted according to the device operating environment, n g is the nonlinear adjustment index, R g The urgency of the fault response;

[0063] Based on the fault response urgency, design targeted fault response measures, including decisions on repair and replacement parts, using the formula:

[0064]

[0065] Get the response strategy score, where δ g and σ g is the weight factor, C g represents the estimated cost of implementing the response plan, π g is a normalization factor used to balance the cost-effectiveness of response measures, R g S is the emergency level of fault response, g represents the response strategy score;

[0066] Based on the response strategy score, list the steps, required resources, time frame and execution team, apply the scoring model, and use the formula:

[0067] P g =∑(S g ·λ g )

[0068] Generate a fault response plan, where λ g is the weight factor adjusted for the differentiated response strategy, S g is the response strategy score, P g Represents a failure response plan.

[0069] As a further solution of the present invention, the steps of obtaining the fault handling feedback result are specifically as follows:

[0070] After implementing the fault response plan, capture key operational data to evaluate the initial effect of fault handling, using the formula:

[0071]

[0072] Get the initial fault assessment score, where F v Denotes the initial fault assessment score, D v Represents the monitored operation data, α v is the adjustment factor for the fault type, β v is the adjustment constant for environmental factors, Z v is the equipment service life parameter, m v is the impact index;

[0073] Based on the preliminary fault assessment score and real-time monitoring data, the assessment score is updated in combination with the real-time data using the formula:

[0074]

[0075] Generate fault adjustment analysis results, where A v represents the analysis result after fault adjustment, ξ v and μ v is the weight factor, S vrepresents real-time status data, ρ v is the normalization factor, κ v is the adjustment factor for time sensitivity, T v Represents the time from the occurrence of the fault to the present, F v Provide a preliminary assessment score for the fault;

[0076] According to the analysis results after the fault adjustment, evaluate the effectiveness of the adjustment measures and the impact on the operating status, and apply the formula:

[0077] R v =max(A v ,θ v +ω v ·Y v )

[0078] Generate fault handling feedback results, where θ v is the basic decision threshold, ω v is the adjustment factor for the target condition, Y v represents the impact of preventive measures, R v Indicates the fault handling feedback result, A v Indicates the analysis result after fault adjustment.

[0079] Compared with the prior art, the advantages and positive effects of the present invention are:

[0080] In the present invention, by intelligently classifying and coding equipment faults, the formation of an automated standardized fault data set using a device driver component improves data consistency and comparability. Automated processing not only reduces the time consumption of data sorting, but also improves the response speed of fault diagnosis. A control circuit is used to intelligently analyze the association rules between fault types and generate a conditional fault probability matrix, providing an accurate mathematical basis for fault prediction, thereby enhancing the accuracy of the prediction. Fault prediction intelligent analysis enhances the ability to identify impending faults through the fusion of vibration sensors and real-time operation data, allowing the maintenance team to take measures before the fault occurs, effectively reducing equipment downtime. The design and implementation of intelligent fault response strategies improves the adaptability and effectiveness of maintenance measures through real-time monitoring and dynamic adjustment of strategies to match the current equipment status, thereby ensuring the maximum operating efficiency of the equipment and extending the service life of the equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] Figure 1 is a system flow chart of the present invention;

[0082] Figure 2 A flow chart of a standardized fault data set in the present invention;

[0083] Figure 3 It is a flow chart of the fault association diagram in the present invention;

[0084] Figure 4 It is a flow chart of the conditional fault probability matrix in the present invention;

[0085] Figure 5 A flowchart for predicting fault types in the present invention;

[0086] Figure 6 A flow chart of the fault impact assessment results in the present invention;

[0087] Figure 7 A flowchart of the fault response plan in the present invention;

[0088] Figure 8 This is a flow chart of the fault handling feedback result in the present invention. DETAILED DESCRIPTION

[0089] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0090] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a target position, be constructed and operated in the target position, and therefore cannot be understood as a limitation on the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0091] See also Figure 1 , the interactive manual fault diagnosis system based on artificial intelligence includes:

[0092] The fault status intelligent coding module intelligently classifies and codes equipment faults based on the device driver component, obtains fault type identification, and converts historical fault data into standardized codes to form a standardized fault data set;

[0093] The intelligent fault correlation analysis module uses standardized fault data sets and intelligently analyzes the correlation rules between fault types through control circuits to generate a fault correlation graph. Based on the fault correlation graph, it calculates the conditional probability of fault occurrence and generates a conditional fault probability matrix.

[0094] The fault prediction intelligent analysis module uses the conditional fault probability matrix to intelligently predict the occurrence of potential faults based on the vibration sensor data and the real-time operation data of the current equipment, generates the predicted fault type, evaluates the impact and urgency of the fault based on the predicted fault type, and generates the fault impact assessment result;

[0095] The intelligent fault response strategy module designs targeted intelligent fault response measures based on the fault impact assessment results, generates a fault response plan, implements the fault response plan and monitors the effects, adjusts measures and matches the real-time situation, and generates fault handling feedback results.

[0096] Fault type identification includes fault code, fault name and fault priority; standardized fault data set includes the number of data entries, entry attributes and data integrity; fault association graph includes the number of nodes, number of edges and node density; conditional fault probability matrix includes matrix dimension, probability interval and dependency; predicted fault type includes fault type, prediction accuracy and risk level; fault impact assessment results include affected systems, downtime and recovery strategy; fault response plan includes countermeasures, resource allocation and expected results; fault handling feedback results include feedback type, processing time and satisfaction evaluation.

[0097] See also Figure 2 ,The specific steps for obtaining the standardized fault data set are:

[0098] Based on the device driver component, the device faults are intelligently classified and coded, the fault codes are located through the index table, and the status information is updated using the formula:

[0099]

[0100] Calculate and generate the weighted fault type index, where H ai represents the identifier of the i-th fault type, W ai represents the weight parameter of the i-th fault type, n a Represents the total number of fault types, T a represents the weighted fault type index;

[0101] The difference between the weighted fault type index and the historical fault data is calculated using the formula:

[0102]

[0103] The standard deviation index is calculated, where H ai Represents the identifier of the i-th fault type, T a represents the weighted fault type index, S a Represents the standard deviation indicator, n a Represents the number of fault types;

[0104] Each fault identification is normalized using the standard deviation indicator, using the formula:

[0105]

[0106] Each fault data is standardized to generate a standardized fault data set, where H ai Represents the identifier of the i-th fault type, T a represents the weighted fault type index, S a represents the standard deviation indicator, D a Represents the normalized fault dataset.

[0107] Calculation of weighted fault type index:

[0108] H ai : Fault type identification, which indicates the specific identification value of a certain type of fault, obtained through the device's sensor monitoring data or recorded data. Its value is based on the fault frequency, severity or classification standard;

[0109] W ai : Weight parameter, which indicates the importance of a certain fault type. The weight parameter is calculated through the analysis of historical fault data or assigned by expert evaluation. Its value is between 0 and 1;

[0110] n a : The number of fault types is the total number of all fault types monitored in the device, which can be obtained through the data records of the device driver components;

[0111] Through the sensor records of the equipment, the type identification values ​​of three types of faults are collected, namely: H a1 =0.8, H a2 =0.5, H a3 =0.7;

[0112] Through historical data analysis, determine the corresponding weight parameters: W a1 =0.3, W a2 =0.4, W a3 =0.3;

[0113] The number of fault types is: n a =3;

[0114] Calculate the weighted fault type index T a :

[0115]

[0116] T a =0.24+0.2+0.21=0.65

[0117] Weighted fault type index T a =0.65, which means the weighted average value of the fault types after comprehensive consideration of the weights.

[0118] Calculation of the standard deviation indicator:

[0119] S a : Standard deviation index, which indicates the deviation between the fault type identification value and the weighted fault type index, reflecting the consistency of fault data;

[0120] The remaining parameters H ai 、T a 、n a Same as the previous step;

[0121] The weighted fault type index T in the previous step a =0.65 into the calculation;

[0122] Calculate each fault type identification value and T a The difference is squared and summed, and then divided by the number of fault types;

[0123] Calculate S a :

[0124]

[0125]

[0126] Standard Deviation Index S a =0.126, indicating the discrete degree of the fault type identification value.

[0127] Standardized fault dataset:

[0128] D a : A single value in the standardized fault data set, indicating the result after the fault type identification is standardized;

[0129] The remaining parameters H ai 、T a , S a The steps are the same as above;

[0130] Calculate the normalized value for each fault type identifier:

[0131]

[0132] Substitute T in the previous step a =0.65 and S a =0.126, calculate the standardized results of each fault type identification:

[0133]

[0134] The individual values ​​in the normalized fault dataset are D a1 ≈1.19, D a2 ≈-1.19, D a3 ≈0.40, the value is used to represent the relative position of different fault types after standardization.

[0135] See also Figure 3 , the specific steps for obtaining the fault association graph are:

[0136] Use the standardized fault data set for preliminary analysis, process the fault type and frequency, and use the formula:

[0137]

[0138] Generate fault impact index, where H bi represents the standardized data of the i-th fault type, μ b is the average value of the fault data, σ b is the standard deviation, I b Indicator indicating the degree of impact of the fault;

[0139] The fault impact index is applied to the intelligent control circuit, the fault correlation potential is calculated, and the nonlinear function is used to process the data formula:

[0140]

[0141] Generate a fault correlation potential index, where L b represents the fault correlation potential index, I b Represents the fault impact index, k b and t b Indicates adjustment parameters;

[0142] According to the fault correlation potential index, the correlation strength between differentiated fault types is calculated by statistical analysis method, using the formula:

[0143] G b =∑L b ·E bi

[0144] Generate a fault correlation graph, where G b represents the fault correlation diagram, L b represents the fault correlation potential index, E bi Indicates the existence of the connection between the i-th fault type.

[0145] Fault impact index:

[0146] H bi: Standardized data of a single fault type, obtained from the fault database. For example, the standardized data of fault type A is 0.75;

[0147] μ b : The average value of all fault data, calculated by taking all H bi The average of the values, set the average value to 0.65;

[0148] σ b : Standard deviation, which indicates the fluctuation or dispersion of data. The standard deviation is set to 0.10;

[0149] Calculation process:

[0150] For fault type A, H bi =0.75, μ b =0.65 and σ b =0.10;

[0151] Calculation of Fault Impact Index I b The formula is:

[0152]

[0153] I b =1.0 means that the impact of fault type A is one unit of standard deviation, which means that it is more significant in the fault data relative to the mean.

[0154] Fault correlation potential indicators:

[0155] k b : Parameter for adjusting sensitivity, set by data analysis, set to 1;

[0156] t b : Threshold parameter, which determines the turning point of the activation function, is set to 0.8.

[0157] Calculation process:

[0158] Known I b =1.0, k b =1,t b =0.8;

[0159] Calculation of Fault Correlation Potential Index L b The formula is:

[0160]

[0161] L b =0.55 indicates that the correlation potential of fault type A is above average, which means that there is a certain potential correlation but it is not the strongest.

[0162] Fault correlation diagram:

[0163] E bi : Indicates the existence of the connection between fault types (0 or 1), according to L b Values ​​above a certain threshold, for example, above 0.5 indicate that there is a connection.

[0164] Calculation process:

[0165] Assume that there are three fault types A, B, and C. b The values ​​are 0.55, 0.65, and 0.45;

[0166] The calculation formula of the fault correlation diagram is:

[0167] G b =∑L b ·E bi =0.55·1+0.65·1+0.45·0=1.20

[0168] The value of the fault correlation diagram G b =1.20 indicates that there is a significant correlation between fault types A and B, while C has a weaker correlation, which helps to understand the fault types and provide information for the prioritization of maintenance and fault prevention measures.

[0169] See also Figure 4 , the specific steps for obtaining the conditional failure probability matrix are:

[0170] Extract the weights of the interconnections between nodes of multiple fault types from the fault association graph and calculate the weighted sum of the edges corresponding to each fault node using the formula:

[0171]

[0172] Generate the weighted sum of the faulty nodes, where E bi represents the existence of the connection between the i-th fault type, P cj represents the probability weight of the connection, d c is the attenuation factor used to adjust the weight influence, W c represents the weighted sum of the faulty nodes;

[0173] According to the weighted sum of the faulty nodes, the weighted normalization formula is applied to calculate the conditional probability of the fault occurring, through the formula:

[0174]

[0175] Generate the conditional probability of failure occurrence, where P fc represents the conditional probability of failure, W c represents the weighted sum of the faulty nodes, α c represents the normalization coefficient;

[0176] Using the conditional probability of fault occurrence, we can identify the dependencies between differentiated faults, determine the prerequisites for each fault to be triggered, and calculate the conditional probability values ​​between multiple faults. The formula is:

[0177] M fc =[P fc1 ,P fc2 ,…,P fcn ]

[0178] Construct the conditional failure probability matrix, where P fcn represents the conditional probability of the target fault type, P fc1 , P fc2 represents the conditional probability of the first and second fault types, M fc Represents the conditional failure probability matrix.

[0179] Weighted sum of failed nodes:

[0180] E bi : Indicates the existence of the connection between the faulty nodes, 0 or 1, determined according to the actual connection status monitored by the system;

[0181] P cj : The probability weight of the connection is obtained through statistical analysis of historical fault data;

[0182] d c : Weight attenuation factor, used to adjust the influence of distant and nearby nodes on the current node. It is a constant determined by data analysis and adjusted according to the specific needs of the system and the distribution of historical fault data;

[0183] Set a faulty node to be connected to three nodes, and the connection existence E bi are 1, 0, 1 respectively, and the connection probability weight P cj are 0.8, 0, and 0.6 respectively, and the weight attenuation factor d c Set to 0.9;

[0184] Use the formula:

[0185]

[0186] calculate:

[0187] W c =1·0.8 0.9 +0.0 0.9 +1 0.6 0.9 ≈0.8 0.9 +0.6 0.9

[0188] ≈0.768+0.543=1.311

[0189] W c =1.311 represents the sum of weighted weights of the faulty node, reflecting the degree of fault correlation between the node and other nodes.

[0190] Conditional probability of failure occurring:

[0191] α c : Normalization coefficient, used to adjust the ratio of the sum of weights of different nodes and determine the sensitivity of normalization, which is determined through analysis;

[0192] Use the result W obtained in the previous step c =1.311, set the sum of weighted weights of other nodes to 2.689, α c Set to 1;

[0193] Using the formula

[0194] Calculate P fc :

[0195]

[0196] P fc =0.328 represents the conditional probability of the faulty node failing, indicating the possibility of the node failing under the current system state.

[0197] Conditional failure probability matrix:

[0198] P fc : The conditional probability of each fault type, obtained from the previous step;

[0199] There are three types of faults, whose conditional probabilities are 0.328 (calculated above), 0.452, and 0.220 respectively;

[0200] Use Formula M fc =[P fc1 ,P fc2 ,…,P fcn ];

[0201] Build M fc = [0.328, 0.452, 0.220];

[0202] M fc It is a conditional fault probability matrix, each element of which reflects the conditional probability of the target fault type occurring, which is used for further fault analysis and prediction.

[0203] See also Figure 5 ,The specific steps for obtaining the predicted fault type are:

[0204] Extract the probability vector associated with the current equipment type from the conditional failure probability matrix, adjusting for the impact of real-time operational data, using the formula:

[0205]

[0206] Generate an adjusted failure probability vector, where M fc represents the conditional failure probability matrix, O dk Represents the influence coefficient of the operating data, c d is a constant used to adjust the effect, r d is the enhancement factor, V d represents the adjusted failure probability vector;

[0207] Based on the adjusted fault probability vector, combined with the vibration sensor data, the formula:

[0208]

[0209] Generate a failure index where β d and γ d is the adjustment coefficient, p d is the nonlinear weighting index of vibration data, δ d is the normalization coefficient, V d represents the adjusted failure probability vector, S d Represents vibration sensor data, I d represents the fault index;

[0210] Using the fault index, the threshold determination method is applied to determine the fault type, through the formula:

[0211] T d =max(I d )>θ d

[0212] If T d Exceeding a predetermined threshold value θ d , then it is identified as the corresponding fault type and the predicted fault type is generated, where I d represents the fault index, T d is the fault type determination threshold, θ d Indicates a predetermined threshold.

[0213] The formula for the adjusted failure probability vector is:

[0214]

[0215] M fc : Failure probability from conditional failure probability matrix, extracted directly from the failure database;

[0216] Odk : Real-time operation data impact coefficient, obtained from the system real-time monitoring data;

[0217] c d : Adjustment constant, used to fine-tune the impact of operational data, determined to be 0.2 based on historical data analysis;

[0218] r d : Enhancement factor, used to amplify the impact of key operating parameters, set to 1.5 based on the system performance adjustment history;

[0219] Set the failure probability matrix M fc For a certain device, it is [0.3, 0.4, 0.3], and the real-time operation data O d The corresponding values ​​are [10, 15, 20];

[0220] Calculate V d :

[0221]

[0222] V d =0.3·(10+0.2) 1.5 +0.4 (15+0.2) 1.5 +0.3 (20 + 0.2) 1.5

[0223] V d =0.3·10.2 1.5 +0.4 15.2 1.5 +0.3 20.2 1.5

[0224] V d =0.3·32.03+0.4·59.29+0.3·90.97≈56.88

[0225] The calculated V d =56.88 represents the adjusted fault probability vector, which reflects the possibility of each fault type under the current operating conditions.

[0226] The formula for the failure index is:

[0227]

[0228] β d , γ d : Weight adjustment coefficients, 0.7 and 0.3 respectively, obtained based on system optimization tests;

[0229] S d : Vibration sensor data, obtained in real time from device sensors;

[0230] pd : Nonlinear weighting index, set to 2, used to strengthen the impact of vibration data;

[0231] δ d : Normalization coefficient, set to 100 to standardize the results;

[0232] Set the vibration sensor data S d is 40;

[0233] Computation I d :

[0234]

[0235] I d =5.198 represents the current comprehensive fault index of the equipment, which is used to evaluate the critical state of the fault.

[0236] The formula for fault type determination threshold is:

[0237] T d =max(I d )>θ d

[0238] θ d : Judgment threshold, set to 5 based on the equipment operation standard;

[0239] Example and calculation process:

[0240] Comparison I d and θ d :

[0241] T d =max(I d )>θ d

[0242] T d =5.198>5

[0243] Judge T d Exceeding the threshold value θ d , identified as a fault occurrence, corresponding to the fault type.

[0244] See also Figure 6 , the specific steps for obtaining the fault impact assessment results are:

[0245] Extract the associated fault impact parameters from the predicted fault type, including production downtime and quality loss, using the formula:

[0246]

[0247] Generate a fault impact score, where F hm is the severity of the fault impact, Ghm is the loss caused by the failure, d h is the adjustment factor for the fault frequency, t h is the nonlinear adjustment coefficient affecting the depth, P h represents the fault impact score;

[0248] Combine the fault impact score with the current equipment maintenance and operating status data, and use the formula:

[0249]

[0250] The fault impact index is obtained, where κ h and μ h are the coefficients for extracting weights from the equipment status and fault impact score, respectively. h Represents the health status of the device, h is the normalization factor, P h is the fault impact score, I h represents the fault impact score;

[0251] According to the fault impact index, a logical judgment method is used to determine the fault impact level and the priority of emergency treatment. h is greater than the set threshold θ h , the result is "high impact", otherwise it is "manageable impact", and the fault impact assessment result is generated.

[0252] Failure Impact Score:

[0253]

[0254] F hm : Indicates the impact level of the target fault type, obtained from the analysis of historical fault data. For example, the level of a certain fault type affecting the production line stoppage is 5 (in the range of 1 to 10);

[0255] G hm : The direct loss caused by the failure, measured in economic losses, for example, USD 10,000;

[0256] d h : The adjustment factor of the fault occurrence frequency is set according to the historical frequency and is set to 0.1;

[0257] t h : Nonlinear adjustment coefficient of impact depth, used to adjust the weighted calculation of impact, set to 2;

[0258] There are two types of faults, each with an impact level of F hm 5 and 7 respectively, with a direct loss of G hm 10,000 and 15,000 USD respectively;

[0259] Calculate P h :

[0260] P h =5·(10000+0.1) 2 +7 (15000+0.1) 2

[0261] P h =5·100000.01+7·225000.01≈1975000.14

[0262] The calculated P h ≈1975000.14 represents the fault impact weight after considering the loss and occurrence frequency, which is used to further evaluate the overall impact of the fault.

[0263] The formula for the Failure Impact Score is:

[0264]

[0265] κ h , μ h : The coefficients of weights extracted from the equipment status and fault impact scores are set to 0.7 and 0.3 respectively;

[0266] H h : The health status of the device, which is obtained by real-time monitoring of the device sensor, and the current value is set to 80 (in the range of 0 to 100);

[0267] ξ h : Normalization factor, used to ensure that the index value is within a reasonable range, set to 1000000;

[0268] Computation I h :

[0269]

[0270] I h =1.382524 represents the current comprehensive fault impact index of the equipment. A lower value indicates that the fault impact is manageable.

[0271] Failure impact assessment results:

[0272] A h : Fault impact assessment results;

[0273] θ h : The set threshold is set to 1.5 and is used to determine the fault impact level;

[0274] Use logical judgment formula to determine the fault impact level: A h =If I h >θh , then “high impact”, otherwise “manageable impact”;

[0275] Specific judgment: A h =If 1.382524>1.5, A h is “manageable impact”.

[0276] See also Figure 7 , the specific steps for obtaining the fault response plan are:

[0277] Analyze the fault impact assessment results to determine the urgency and priority of the fault response using the formula:

[0278]

[0279] Get the fault response urgency, where E g Indicates the fault impact level, γ g is a coefficient adjusted based on device category and impact level, ∈ g is a fixed value adjusted according to the device operating environment, n g is the nonlinear adjustment index, R g The urgency of the fault response;

[0280] Based on the urgency of the fault response, design targeted fault response measures, including decisions on repairs and replacement parts, using the formula:

[0281]

[0282] Get the response strategy score, where δ g and σ g is the weight factor, C g represents the estimated cost of implementing the response plan, π g is a normalization factor used to balance the cost-effectiveness of response measures, R g S is the emergency level of fault response, g represents the response strategy score;

[0283] Based on the response strategy score, list the steps, required resources, time frame and execution team, apply the scoring model, and use the formula:

[0284] P g =∑(S g ·λ g )

[0285] Generate a fault response plan, where λ g is the weight factor adjusted for the differentiated response strategy, S g is the response strategy score, P g Represents a failure response plan.

[0286] The formula for fault response urgency is:

[0287]

[0288] E g : Fault impact level score, based on the fault impact assessment result, for example, the fault impact assessment score is 75 (out of 100);

[0289] γ g : Equipment type and impact level adjustment factor, derived from historical data analysis, set to 1.2;

[0290] ∈ g : Fixed value for equipment operating environment adjustment, obtained from the equipment target operating condition data, set to 5;

[0291] n g : nonlinear adjustment index, used to strengthen the response to high-impact faults, set to 1.5;

[0292] Calculate the urgency of the fault response and calculate R g :

[0293] R g =1.2·(75+5) 1.5 =1.2·80 1.5

[0294] R g =1.2·714.14=856.97

[0295] This result indicates that the urgency of the fault response is very high based on the fault's impact level score and the target condition of the equipment.

[0296] The formula for the response strategy score is:

[0297]

[0298] δ g and σ g : Weight factor, representing the importance of fault urgency and cost-effectiveness, set to 0.6 and 0.4 respectively;

[0299] C g : The estimated cost of the response plan, for example, the cost estimate is $5,000;

[0300] π g : Normalization factor, to ensure that the final response strategy score is within a reasonable range, set to 10000;

[0301] Calculate the response strategy score S g :

[0302]

[0303] This score is used to assess the overall effectiveness of the response plan, including cost-effectiveness and urgency.

[0304] Formula for Failure Response Plan:

[0305] P g =∑(S g ·λ g )

[0306] λ g : Policy weight factor, which is set according to the priority of the response and resource availability and is set to 0.8;

[0307] Generate a fault response plan using the comprehensive scoring model and calculate P g :

[0308] P g =0.2514·0.8=0.20112

[0309] This combined score determines the final execution priority of the response plan, with high scores representing urgent and high-priority response measures.

[0310] See also Figure 8 , the specific steps for obtaining the fault handling feedback results are as follows:

[0311] After implementing the fault response plan, capture key operational data to evaluate the initial effect of fault handling, using the formula:

[0312]

[0313] Get the initial fault assessment score, where F v Denotes the initial fault assessment score, D v Represents the monitored operation data, α v is the adjustment factor for the fault type, β v is the adjustment constant for environmental factors, Z v is the equipment service life parameter, m v is the impact index;

[0314] Based on the preliminary fault assessment score and real-time monitoring data, the assessment score is updated in combination with real-time data using the formula:

[0315]

[0316] Generate fault adjustment analysis results, where A v represents the analysis result after fault adjustment, ξ v and μ v is the weight factor, S vrepresents real-time status data, ρ v is the normalization factor, κ v is the adjustment factor for time sensitivity, T v Represents the time from the occurrence of the fault to the present, F v Provide a preliminary assessment score for the fault;

[0317] According to the analysis results after fault adjustment, evaluate the effectiveness of the adjustment measures and their impact on the operating status, and apply the formula:

[0318] R v =max(A v ,θ v +ω v ·Y v )

[0319] Generate fault handling feedback results, where θ v is the basic decision threshold, ω v is the adjustment factor for the target condition, Y v represents the impact of preventive measures, R v Indicates the fault handling feedback result, A v Indicates the analysis result after fault adjustment.

[0320] The formula for the initial fault assessment score is:

[0321]

[0322] α v : Fault type adjustment factor, determined through historical data analysis, set to 1.5;

[0323] D v : Monitored operation data, set to 85 during evaluation;

[0324] β v : The adjustment constant of environmental factors is obtained through analysis of equipment location and environmental conditions and is set to 0.05;

[0325] Z v : Equipment service life, reflecting the impact of equipment wear on the response effect, set to 10 years;

[0326] m v : Impact index, strengthens the impact of the target operation, set to 2;

[0327] Calculate F v :

[0328] F v =1.5·(85+0.05·10) 2

[0329] Fv =1.5·(85+0.5) 2 =1.5·85.5 2

[0330] F v =1.5·7310.25=10965.375

[0331] This score reflects a quantitative assessment of the initial effectiveness of the fault handling, with a high score indicating a good initial response.

[0332] The formula for the analysis result after fault adjustment is:

[0333]

[0334] ξ v and μ v : Weight factors, set to 0.7 and 0.3 respectively;

[0335] S v : Real-time status data, set the score to 200;

[0336] ρ v : Normalization factor, set to 10000;

[0337] κ v : Adjustment factor for time sensitivity, set to 0.05;

[0338] T v : The time from the fault occurrence to the present, set to 100 hours;

[0339] Calculate A v :

[0340]

[0341] This result is used to assess whether further adjustments to the response plan are needed, with higher values ​​indicating a need for significant adjustments.

[0342] The formula for fault handling feedback result is:

[0343] R v =max(A v ,θ v +ω v ·Y v )

[0344] θ v : Basic decision threshold, set to 8000;

[0345] ω v : The adjustment coefficient of the target condition is set to 1.1;

[0346] Yv : The impact of preventive measures, set to 50;

[0347] Calculate R v :

[0348] R v =max(8457.28,8000+1.1·50)

[0349] R v =max(8457.28,8000+55)=max(8457.28,8055)=8457.28

[0350] The final score is used to determine the priority and effectiveness of the fault handling strategy to ensure the best response and the minimum impact of the fault.

[0351] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. An interactive manual fault diagnosis system based on artificial intelligence, characterized in that: The system comprises: The fault status intelligent coding module intelligently classifies and codes equipment faults based on the device driver component, obtains fault type identification, and converts historical fault data into standardized codes to form a standardized fault data set; The intelligent fault correlation analysis module uses the standardized fault data set to intelligently analyze the correlation rules between fault types through the control circuit, generates a fault correlation diagram, calculates the conditional probability of fault occurrence based on the fault correlation diagram, and generates a conditional fault probability matrix; The fault prediction intelligent analysis module applies the conditional fault probability matrix, based on the vibration sensor data, combined with the real-time operation data of the current equipment, intelligently predicts the occurrence of potential faults, generates predicted fault types, evaluates the impact and urgency of the faults based on the predicted fault types, and generates fault impact assessment results; The intelligent fault response strategy module designs targeted intelligent fault response measures according to the fault impact assessment results, generates a fault response plan, implements the fault response plan and monitors the effects, adjusts measures and matches the real-time situation, and generates fault handling feedback results.

2. The interactive manual fault diagnosis system based on artificial intelligence according to claim 1 is characterized in that: The steps for obtaining the standardized fault data set are specifically as follows: Based on the device driver component, the device faults are intelligently classified and coded, the fault codes are located through the index table, and the status information is updated using the formula: The weighted fault type index is calculated and generated, where H ai represents the identifier of the i-th fault type, W ai represents the weight parameter of the i-th fault type, n a Represents the total number of fault types, T a represents the weighted fault type index; The weighted fault type index is calculated with the historical fault data using the formula: The standard deviation index is calculated, where H ai Represents the identifier of the i-th fault type, T a represents the weighted fault type index, S a Represents the standard deviation indicator, n a Represents the number of fault types; Each fault identification is normalized using the standard deviation indicator, using the formula: Each fault data is standardized to generate a standardized fault data set, where H ai Represents the identifier of the i-th fault type, T a represents the weighted fault type index, S a represents the standard deviation indicator, D a Represents the normalized fault dataset.

3. The interactive manual fault diagnosis system based on artificial intelligence according to claim 2 is characterized in that: The steps of obtaining the fault association diagram are specifically as follows: The standardized fault data set is used for preliminary analysis, and the fault type and frequency are processed. The formula is: Generate fault impact index, where H bi represents the standardized data of the i-th fault type, μ b is the average value of the fault data, σ b is the standard deviation, I b Indicator indicating the degree of impact of the fault; The fault impact index is applied to the intelligent control circuit, the fault correlation potential is calculated, and the nonlinear function is used to process the data formula: Generate a fault correlation potential index, where L b represents the fault correlation potential index, I b Represents the fault impact index, k b and t b Indicates adjustment parameters; According to the fault correlation potential index, the correlation strength between the differentiated fault types is calculated by statistical analysis method, using the formula: G b =∑L b ·E bi Generate a fault correlation graph, where G b represents the fault correlation diagram, L b represents the fault correlation potential index, E bi Indicates the existence of the connection between the i-th fault type.

4. The interactive manual fault diagnosis system based on artificial intelligence according to claim 3 is characterized in that: The steps for obtaining the conditional failure probability matrix are specifically as follows: The weights of the interconnections between nodes of multiple fault types are extracted from the fault association graph, and the weighted sum of the edges corresponding to each fault node is calculated using the formula: Generate the weighted sum of the faulty nodes, where E bi represents the existence of the connection between the i-th fault type, P cj represents the probability weight of the connection, d c is the attenuation factor used to adjust the weight influence, W c represents the weighted sum of the faulty nodes; According to the weighted sum of the fault nodes, the weighted normalization formula is applied to calculate the conditional probability of the fault occurring, through the formula: Generate the conditional probability of failure occurrence, where P fc represents the conditional probability of failure, W c represents the weighted sum of the faulty nodes, α c represents the normalization coefficient; Using the conditional probability of the occurrence of the fault, the dependency between the differentiated faults is identified, the prerequisite for each fault to be triggered is determined, and the conditional probability values ​​between multiple faults are calculated, through the formula: M fc =[P fc1 ,P fc2 ,…,P fcn ] Construct the conditional failure probability matrix, where P fcn represents the conditional probability of the target fault type, P fc1 , P fc2 represents the conditional probability of the first and second fault types, M fc Represents the conditional failure probability matrix.

5. The interactive manual fault diagnosis system based on artificial intelligence according to claim 4 is characterized in that: The steps for obtaining the predicted fault type are specifically as follows: The probability vector associated with the current equipment type is extracted from the conditional failure probability matrix and adjusted for the impact of real-time operational data using the formula: Generate an adjusted failure probability vector, where M fc represents the conditional failure probability matrix, O dk Represents the influence coefficient of the operating data, c d is a constant used to adjust the effect, r d is the enhancement factor, V d represents the adjusted failure probability vector; Based on the adjusted fault probability vector and combined with the vibration sensor data, the formula is: Generate a failure index where β d and γ d is the adjustment coefficient, p d is the nonlinear weighting index of vibration data, δ d is the normalization coefficient, V d represents the adjusted failure probability vector, S d Represents vibration sensor data, I d represents the fault index; Using the fault index, the threshold determination method is applied to determine the fault type, through the formula: T d =max(I d )>θ d If T d Exceeding a predetermined threshold value θ d , then it is identified as the corresponding fault type and the predicted fault type is generated, where I d represents the fault index, T d is the fault type determination threshold, θ d Indicates a predetermined threshold.

6. The interactive manual fault diagnosis system based on artificial intelligence according to claim 5 is characterized in that: The steps for obtaining the fault impact assessment result are specifically as follows: From the predicted fault types, associated fault impact parameters, including production downtime and quality loss, are extracted using the formula: Generate a fault impact score, where F hm is the severity of the fault impact, G hm is the loss caused by the failure, d h is the adjustment factor for the fault frequency, t h is the nonlinear adjustment coefficient affecting the depth, P h represents the fault impact score; The fault impact score is combined with the current equipment maintenance and operation status data, through the formula: The fault impact index is obtained, where κ h and μ h are the coefficients for extracting weights from the equipment status and fault impact score, respectively. h Represents the health status of the device, h is the normalization factor, P h is the fault impact score, I h represents the fault impact score; According to the fault impact index, a logical judgment method is used to determine the fault impact level and the priority of emergency treatment. h is greater than the set threshold θ h , the result is "high impact", otherwise it is "manageable impact", and the fault impact assessment result is generated.

7. The interactive manual fault diagnosis system based on artificial intelligence according to claim 6 is characterized in that: The steps for obtaining the fault response plan are specifically as follows: Analyze the fault impact assessment results to determine the urgency and priority of the fault response using the formula: Get the fault response urgency, where E g Indicates the fault impact level, γ g is a coefficient adjusted based on device category and impact level, ∈ g is a fixed value adjusted according to the device operating environment, n g is the nonlinear adjustment index, R g The urgency of the fault response; Based on the fault response urgency, design targeted fault response measures, including decisions on repair and replacement parts, using the formula: Get the response strategy score, where δ g and σ g is the weight factor, C g represents the estimated cost of implementing the response plan, π g is a normalization factor used to balance the cost-effectiveness of response measures, R g S is the emergency level of fault response, g represents the response strategy score; Based on the response strategy score, list the steps, required resources, time frame and execution team, apply the scoring model, and use the formula: P g =∑(S g ·l g ) Generate a fault response plan, where λ g is the weight factor adjusted for the differentiated response strategy, S g is the response strategy score, P g Represents a failure response plan.

8. The interactive manual fault diagnosis system based on artificial intelligence according to claim 7 is characterized in that: The steps for obtaining the fault handling feedback result are specifically as follows: After implementing the fault response plan, capture key operational data to evaluate the initial effect of fault handling, using the formula: Get the initial fault assessment score, where F v Denotes the initial fault assessment score, D v Represents the monitored operation data, α v is the adjustment factor for the fault type, β v is the adjustment constant for environmental factors, Z v is the equipment service life parameter, m v is the impact index; Based on the preliminary fault assessment score and real-time monitoring data, the assessment score is updated in combination with the real-time data using the formula: Generate fault adjustment analysis results, where A v represents the analysis result after fault adjustment, ξ v and μ v is the weight factor, S v represents real-time status data, ρ v is the normalization factor, κ v is the adjustment factor for time sensitivity, T v Represents the time from the occurrence of the fault to the present, F v Provide a preliminary assessment score for the fault; According to the analysis results after the fault adjustment, evaluate the effectiveness of the adjustment measures and the impact on the operating status, and apply the formula: R v =max(A v ,i v +oh v ·Y v ) Generate fault handling feedback results, where θ v is the basic decision threshold, ω v is the adjustment factor for the target condition, Y v represents the impact of preventive measures, R v Indicates the fault handling feedback result, A v Indicates the analysis result after fault adjustment.

Images