Fault detection method in 3D printing

Abstract

The invention discloses a fault detection method in 3D printing, and relates to the technical field of printing fault detection.The method comprises historical fault data analysis and real-time state data detection.The method comprises the steps that historical fault data of a 3D printer is collected firstly, and the maintenance weight is determined through an analytic hierarchy process in combination with multi-dimensional state parameters and fault data; then objective weights are determined based on data discrete characteristics through an entropy weight method, final weights of state parameters are obtained through weighted fusion, and then a fault monitoring scheme containing a state collection scheme and a state analysis scheme is formulated; according to the method, the scientificity of importance judgment of the state parameters is improved through double-weight fusion, the accuracy and timeliness of fault detection are improved through differential collection and multi-dimensional data analysis, the rate of failure report and false report of 3D printing faults is effectively reduced, and the accuracy and timeliness of fault detection are improved. The printing process stability is ensured.

Description

Technical Field

[0001] This invention relates to the field of printing fault detection technology, and more specifically to a fault detection method in 3D printing. Background Technology

[0002] 3D printing technology, as a core technology of additive manufacturing, is widely used in aerospace, medical, and industrial manufacturing fields due to its advantages such as personalized customization and complex structure forming. However, the 3D printing process involves the coordinated control of multiple state parameters, such as nozzle temperature, filament balance, and platform lifting speed. Abnormalities in any parameter can lead to faults such as nozzle clogging, filament breakage, and layer defects, not only resulting in scrapped printed parts but also, in severe cases, damaging the printer's core components and causing economic losses. Therefore, accurate and timely fault detection is crucial to ensuring the quality and efficiency of 3D printing, thus requiring a fault detection method in 3D printing.

[0003] Existing 3D printing fault detection technologies have the following shortcomings: First, the method of determining the weight of state parameters is singular, or it relies solely on subjective experience or is calculated based on a single data feature. This fails to fully reflect the correlation between parameters and faults, causing the monitoring focus to deviate from the core parameters and resulting in missed detection of critical faults.

[0004] Secondly, the status data acquisition scheme lacks specificity. Using a uniform acquisition cycle to monitor all parameters wastes hardware resources and may delay fault detection due to insufficient acquisition frequency of core parameters.

[0005] Third, the data analysis is too narrow in scope, focusing only on anomalies in a single data collection and ignoring trend anomalies in parameters over a preset time period, which makes it impossible to identify periodic cumulative faults in a timely manner.

[0006] Fourth, the early warning logic is simple, relying solely on a single threshold to determine whether to issue an early warning. This makes it susceptible to false alarms due to instantaneous fluctuations, or to confusion in early warning priorities due to the lack of consideration for differences in parameter weights. Summary of the Invention

[0007] To address the aforementioned technical shortcomings, the present invention aims to provide a fault detection method in 3D printing.

[0008] To solve the above technical problems, the present invention adopts the following technical solution: The present invention provides a fault detection method in 3D printing, including the following steps: Step 1, historical fault data analysis: collect historical fault data of 3D printers, analyze the historical fault data of 3D printers, and obtain a 3D printer fault monitoring scheme. Step 2: Real-time status data detection: Based on the 3D printer fault monitoring solution, collect the current status data of the 3D printer, analyze the current status data of the 3D printer, and provide status warnings for the 3D printer.

[0009] Preferably, the analysis of the historical fault records of the 3D printer is carried out in the following specific process: first, the maintenance weights of various state parameters are determined by the analytic hierarchy process, then the objective weights of various state parameters are determined by the entropy weight method, and finally, the final weight value is obtained by weighted fusion of the maintenance weights and objective weights of various state parameters.

[0010] Based on the final weight values ​​of each core associated status parameter, a 3D printer fault monitoring scheme is set up, which includes a 3D printer status acquisition scheme and a 3D printer status analysis scheme.

[0011] Preferably, the analytic hierarchy process is as follows: the historical fault data of the 3D printer includes the abnormal occurrence rate of various status parameters, the average value of the out-of-limit amplitude of various status parameters, the average value of the out-of-limit duration of various status parameters, the average value of the parameter fluctuation amplitude of various status parameters, the number of repairs for various faults, the maximum repair frequency of various faults, the average repair time of various faults, and the downtime impact time of various faults.

[0012] Obtain the anomaly occurrence rate weight factor, the average over-limit amplitude weight factor, the average over-limit duration weight factor, and the average parameter fluctuation amplitude weight factor from the database. Divide the anomaly occurrence rate, the average over-limit amplitude, the average over-limit duration, and the average parameter fluctuation amplitude of each type of state parameter by the corresponding preset threshold, then multiply by the corresponding weight factor, and finally add them together to obtain the anomaly index of each type of state parameter.

[0013] The system retrieves the maintenance frequency weight factor, maximum maintenance frequency weight factor, average maintenance time weight factor, and downtime impact weight factor from the database. It then divides the maintenance frequency, maximum maintenance frequency, average maintenance time, and downtime impact of each type of fault by the corresponding preset threshold, multiplies them by the corresponding weight factor, and finally sums them to obtain the anomaly index of each type of fault.

[0014] Based on the anomaly indices of various state parameters and various faults, correlation analysis is performed on various state parameters and various faults, and the maintenance weights of various state parameters are calculated.

[0015] Preferably, the correlation analysis is performed as follows: using the fault index of a certain type of fault as the set of numerical content, and arranging them sequentially according to the statistical sample order of historical faults, a fault index sequence of that type of fault is obtained, which is recorded as the baseline sequence of that type of fault. Based on the statistical sample order of historical faults of that type of fault, the abnormal index of a certain type of state parameter at each statistical time of historical faults of that type of fault is obtained, and arranged sequentially to obtain the comparison sequence of that type of state parameter of that type of fault.

[0016] The baseline sequence of this type of fault and the corresponding state parameter sequence to be compared are normalized to obtain the standard baseline sequence of this type of fault and the standard comparison sequence of the corresponding state parameter. The absolute difference between the baseline sequence value and the comparison sequence value under the same historical sample is calculated one by one, and the differences of each sample are integrated to form a difference set.

[0017] The minimum and maximum differences are selected from the set of differences. Then, a preset fixed resolution coefficient is obtained from the database. The minimum difference is added to the product of the maximum difference and the fixed resolution coefficient to obtain the first-level correlation. The difference of each sample is added to the product of the maximum difference and the fixed resolution coefficient to obtain the first-level correlation of each sample. Then, the first-level correlation is divided by the first-level correlation of each sample to obtain the correlation coefficient of each sample. The arithmetic mean is calculated to obtain the final correlation. This gives the final correlation of the state parameters of this type of fault, and then the final correlation of the state parameters of various types of faults.

[0018] The beneficial effects of this invention are as follows: 1. This invention first collects historical fault data of 3D printers, determines maintenance weights by combining multi-dimensional state parameters and fault data through the analytic hierarchy process, and then determines objective weights based on the discrete characteristics of data through the entropy weight method. The weighted fusion is then used to obtain the final weights of the state parameters, thereby formulating a fault monitoring scheme that includes a state acquisition scheme and a state analysis scheme. Subsequently, based on this scheme, the current state data of the 3D printer is collected to provide a state warning. This invention improves the scientific nature of the determination of the importance of state parameters through dual-weight fusion, and improves the accuracy and timeliness of fault detection through differentiated acquisition and multi-dimensional data analysis, effectively reducing the false alarm rate and missed alarm rate of 3D printing faults, and ensuring the stability of the printing process.

[0019] 2. This invention innovatively employs a weighted fusion of the Analytic Hierarchy Process (AHP) and the Entropy Weight Method to determine the final weights of state parameters. The AHP combines multi-dimensional parameter data such as the anomaly rate and over-limit amplitude of state parameters, as well as fault data such as the number of repairs and downtime duration, to comprehensively calculate the maintenance weights, fully considering the importance of parameters from a fault maintenance perspective. The Entropy Weight Method calculates objective weights based on the dispersion of historical data, eliminating the interference of subjective experience. This dual-weight fusion achieves complementarity between subjective experience and objective data, making the determination of the importance of state parameters more comprehensive and scientific, and providing a core basis for subsequent accurate monitoring.

[0020] 3. This invention formulates a differentiated data acquisition scheme based on the final weight of state parameters. For key state parameters with higher weights, the acquisition cycle is shortened through an acquisition cycle correction factor, thereby increasing the monitoring frequency. For non-key parameters, a basic acquisition cycle is used to avoid resource waste. This scheme ensures real-time monitoring of core parameters while also taking into account the rational allocation of system resources, effectively improving the targeting and efficiency of state data acquisition.

[0021] 4. In the real-time data analysis phase, this invention simultaneously considers both the "detection anomalies" of a single data acquisition and the "periodic anomalies" of the data average over a preset time period, achieving dual coverage of both instantaneous sudden faults and trend-based cumulative faults. Compared to existing single-dimensional analysis techniques, this invention can more comprehensively capture various fault precursors, significantly reducing the risk of missed fault detection, and is particularly suitable for detecting periodic cumulative faults in the 3D printing process.

[0022] 5. The early warning logic of this invention combines a fault detection index and a periodic anomaly dual judgment standard: the fault detection index is calculated by weighting the anomaly rate of abnormal data with the corresponding parameter weights, fully considering the differences in the importance of different parameters and avoiding the one-sidedness of a single threshold judgment; at the same time, the direct early warning mechanism for periodic anomalies ensures timely response to trending faults. This dual early warning logic not only improves the accuracy of early warnings but also clarifies the early warning priority, effectively reducing false alarms and providing clear guidance for staff to quickly handle faults. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a schematic diagram of the implementation steps of the method of the present invention. Detailed Implementation

[0025] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] according to Figure 1 As shown, the present invention provides a fault detection method in 3D printing, including the following steps: Step 1, historical fault data analysis: collect historical fault data of 3D printers, analyze the historical fault data of 3D printers, and obtain a 3D printer fault monitoring scheme.

[0027] In one specific embodiment, the historical fault data collection process is as follows: The historical fault data of the 3D printer includes the abnormal occurrence rate of various status parameters, the average value of the out-of-limit amplitude of various status parameters, the average value of the out-of-limit duration of various status parameters, the average value of the parameter fluctuation amplitude of various status parameters, the number of repairs for various faults, the maximum repair frequency of various faults, the average repair time of various faults, and the downtime impact time of various faults.

[0028] Staff preset thresholds for various status parameters. When any status parameter exceeds the threshold, it is recorded as an anomaly. The number of anomalies is counted and divided by the printing duration to obtain the anomaly rate. When an anomaly occurs, the anomaly value is subtracted from the corresponding threshold and then divided by the corresponding threshold. After absolute value conversion, the over-limit range of each anomaly is obtained. The average over-limit range is calculated. When an anomaly occurs, the duration of each anomaly is counted and the average over-limit duration is calculated. The average parameter fluctuation range is obtained by subtracting the minimum value from the maximum value of each status parameter.

[0029] When printing faults occur, the number of printing faults is counted to obtain the number of repairs for each type of fault. The number of repairs within each time period is counted and divided by the statistical duration to obtain the repair frequency, and then the maximum repair frequency is obtained. When repairing printing faults, the timestamps of the printing fault occurrence, repair occurrence, repair completion, and repair continuation are recorded. Based on the timestamps of the printing fault occurrence and repair completion, the repair duration of each repair is obtained, and the average repair duration is calculated from the average. Based on the timestamps of the printing fault occurrence and repair continuation, the downtime impact duration of each type of fault is obtained.

[0030] It should be noted that the types of status data include, but are not limited to, printhead temperature, resin powder filament balance, platform lifting speed, extruder speed, 3D printing resolution parameters, and filament usage time.

[0031] The types of malfunctions include, but are not limited to, nozzle clogging, broken wires, and layering defects.

[0032] In one specific embodiment, the analysis of the historical fault records of the 3D printer is carried out in the following process: first, the maintenance weights of various state parameters are determined by the analytic hierarchy process (AHP); then, the objective weights of various state parameters are determined by the entropy weight method; finally, the final weight value is obtained by weighted fusion of the maintenance weights and objective weights of various state parameters.

[0033] Based on the final weight values ​​of each core associated status parameter, a 3D printer fault monitoring scheme is set up, which includes a 3D printer status acquisition scheme and a 3D printer status analysis scheme.

[0034] In one specific embodiment, the analytic hierarchy process is as follows: obtain the anomaly occurrence rate weight factor, the average over-limit amplitude weight factor, the average over-limit duration weight factor, and the average parameter fluctuation amplitude weight factor from the database; divide the anomaly occurrence rate, average over-limit amplitude, average over-limit duration, and average parameter fluctuation amplitude of each type of state parameter by the corresponding preset threshold, then multiply by the corresponding weight factor, and finally add them together to obtain the anomaly index of each type of state parameter.

[0035] The system retrieves the maintenance frequency weight factor, maximum maintenance frequency weight factor, average maintenance time weight factor, and downtime impact weight factor from the database. It then divides the maintenance frequency, maximum maintenance frequency, average maintenance time, and downtime impact of each type of fault by the corresponding preset threshold, multiplies them by the corresponding weight factor, and finally sums them to obtain the anomaly index of each type of fault.

[0036] Based on the anomaly indices of various state parameters and various faults, correlation analysis is performed on various state parameters and various faults, and the maintenance weights of various state parameters are calculated.

[0037] In one specific embodiment, the correlation analysis is performed as follows: using the fault index of a certain type of fault as the set of numerical content, and arranging them sequentially according to the statistical sample order of historical faults, a fault index sequence of that type of fault is obtained, which is denoted as the baseline sequence of that type of fault. Based on the statistical sample order of historical faults of that type of fault, the abnormal index of a certain type of state parameter at each statistical time of historical faults of that type of fault is obtained, and arranged sequentially to obtain the comparison sequence of that type of state parameter of that type of fault.

[0038] The baseline sequence of this type of fault and the corresponding state parameter sequence to be compared are normalized to obtain the standard baseline sequence of this type of fault and the standard comparison sequence of the corresponding state parameter. The absolute difference between the baseline sequence value and the comparison sequence value under the same historical sample is calculated one by one, and the differences of each sample are integrated to form a difference set.

[0039] The minimum and maximum differences are selected from the set of differences. Then, a preset fixed resolution coefficient is obtained from the database. The minimum difference is added to the product of the maximum difference and the fixed resolution coefficient to obtain the first-level correlation. The difference of each sample is added to the product of the maximum difference and the fixed resolution coefficient to obtain the first-level correlation of each sample. Then, the first-level correlation is divided by the first-level correlation of each sample to obtain the correlation coefficient of each sample. The arithmetic mean is calculated to obtain the final correlation. This gives the final correlation of the state parameters of this type of fault, and then the final correlation of the state parameters of various types of faults.

[0040] In one specific embodiment, the calculation of maintenance weights for various status parameters is carried out as follows: The various status parameters of various faults are sorted in descending order of final correlation degree to obtain a sequence of various status parameters for each type of fault. The first preset number of various status parameters in the sequence of various status parameters for each type of fault are recorded as core related status parameters. The final correlation degrees of the core related status parameters for each type of fault are added together to obtain the weight base of each type of fault. The final correlation degree of each status parameter for each type of fault is divided by the corresponding weight base to obtain the weight factor of each status parameter for each type of fault.

[0041] The abnormality index of each state parameter of each type of fault is multiplied by the corresponding weight factor, and then summed to obtain the detection fault index of each type of fault. The detection fault index and abnormality index of each type of fault are multiplied by the corresponding threshold, then multiplied by the corresponding weight factor, and finally summed to obtain the monitoring index of each type of fault.

[0042] Based on the core associated status parameters of various faults, the core faults corresponding to each status parameter are obtained. The maximum monitoring index of each core fault corresponding to each status parameter is recorded as the monitoring index of each status parameter. The monitoring index of each status parameter is obtained by summing the monitoring indices of each status parameter. The maintenance weight of each status parameter is obtained by dividing the monitoring index of each status parameter by the total monitoring index.

[0043] In one specific embodiment, the entropy weight method is analyzed as follows: the values ​​of various state parameters corresponding to the occurrence of historical faults with a preset number of faults are summarized to obtain data groups of various state parameters, and all data groups of state parameters are integrated into a total dataset to be calculated.

[0044] Divide the total dataset by the corresponding threshold to obtain the normalized total dataset. Divide each individual data point under each parameter class by the sum of all data points under that parameter class to obtain the proportion of each data point. This gives the proportion of each data point for each type of state parameter. Substituting the proportions of each data point for various state parameters into the formula for calculating information entropy, we obtain the information entropy values ​​for each state parameter.

[0045] The formula for calculating information entropy is: ,in Let be the information entropy value of the i-th type of state parameter, where i is the type number of the state parameter, and ln is the natural logarithm. Let j represent the percentage of the j-th data point under the i-th state parameter category out of the total sum of all data in that category. j is the data content number, j=1,2……n, and n is a positive integer greater than 2.

[0046] The total information entropy value is obtained by summing the information entropy values ​​of various state parameters. The objective weight of each state parameter is obtained by dividing the information entropy value of each state parameter by the total information entropy value. The weight factor of maintenance weight and the weight factor of objective weight are obtained from the database. The maintenance weight and objective weight of each state parameter are multiplied by the corresponding weight factor, and then summed to obtain the final weight value of each state parameter.

[0047] In one specific embodiment, the 3D printer status acquisition scheme is implemented as follows: The operator presets the basic acquisition cycle corresponding to various status parameters, and sorts the various status parameters in descending order of their final weight values ​​to obtain a status parameter acquisition sequence. A preset number of various status parameters at the beginning of the status parameter acquisition sequence are designated as key status parameters. The operator presets an acquisition cycle correction factor for each position in the sequence, and multiplies the basic acquisition cycle corresponding to each status parameter by the corresponding acquisition cycle correction factor to obtain the corrected acquisition cycle corresponding to each status parameter. The 3D printer status acquisition scheme is as follows: data is acquired for each type of status parameter according to the corresponding corrected acquisition cycle.

[0048] In one specific embodiment, the 3D printer status analysis scheme is implemented as follows: based on the final weight values ​​of various status parameters, when abnormal status parameters occur, various abnormal status parameters are weighted and analyzed according to their corresponding final weight values.

[0049] Step 2: Real-time status data detection: Based on the 3D printer fault monitoring solution, collect the current status data of the 3D printer, analyze the current status data of the 3D printer, and provide status warnings for the 3D printer.

[0050] In one specific embodiment, the analysis of the current 3D printer status data is as follows: the current 3D printer status data includes various types of status data collected each time within the current preset time period; The 3D printer status acquisition scheme collects various types of status data from each acquisition within the current preset time period. If a certain type of status data collected within the current preset time period is greater than the corresponding preset first threshold, the status data of that type collected within the current preset time period is recorded as abnormal data, thus obtaining the current types of abnormal data.

[0051] The average value of each type of state data collected within the current preset time period is obtained by averaging the data. If the average value of a certain type of state data within the current preset time period is greater than the corresponding preset second threshold, it is recorded as periodic abnormal data. In this way, various types of periodic abnormal data and various types of detection abnormal data within the current preset time period are obtained.

[0052] In one specific embodiment, the status warning process for the 3D printer is as follows: The final weight values ​​corresponding to various types of abnormal detection data within the current preset time period are added together to obtain the current total final weight value. The values ​​of various types of abnormal detection data are subtracted from the corresponding threshold values, and then divided by the corresponding threshold values. After absolute value conversion, the abnormality rate of various types of abnormal detection data is obtained. The abnormality rate of various types of abnormal detection data is multiplied by the corresponding final weight value, and then divided by the current total final weight value to obtain the current fault detection index. If the current fault detection index is greater than the preset current fault detection index threshold, a warning is issued.

[0053] At the same time, if a certain type of periodic abnormal data appears within the current preset time period, an early warning will be issued directly.

[0054] The database stores weight factors for anomaly occurrence rate, average over-limit amplitude, average over-limit duration, average parameter fluctuation amplitude, number of repairs, maximum repair frequency, average repair duration, downtime impact duration, preset fixed resolution coefficients, repair weights, and objective weights.

[0055] The examples described in this invention are not limited to the specific embodiments listed above. The examples are merely illustrative to facilitate understanding of the invention and do not constitute a limitation on the scope of protection of this invention. Any modifications, equivalent substitutions, etc., made within the spirit and principles of this invention should be included within the scope of protection.

[0056] The above description is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined in this specification, they should all fall within the protection scope of the present invention.

Claims

1. A fault detection method in 3D printing, characterized in that, Includes the following steps: Step 1: Historical Fault Data Analysis: Collect historical fault data of 3D printers, analyze the historical fault data of 3D printers, and obtain a 3D printer fault monitoring solution. Step 2: Real-time status data detection: Based on the 3D printer fault monitoring solution, collect the current status data of the 3D printer, analyze the current status data of the 3D printer, and provide status warnings for the 3D printer.

2. The fault detection method in 3D printing according to claim 1, characterized in that, The analysis of the historical fault records of the 3D printer is carried out in the following specific process: First, the maintenance weights of various state parameters are determined by the analytic hierarchy process (AHP). Then, the objective weights of various state parameters are determined by the entropy weight method. Finally, the final weight values ​​are obtained by weighted fusion of the maintenance weights and objective weights of various state parameters. Based on the final weight values ​​of each core associated status parameter, a 3D printer fault monitoring scheme is set up, which includes a 3D printer status acquisition scheme and a 3D printer status analysis scheme.

3. The fault detection method in 3D printing according to claim 2, characterized in that, The specific analysis process of the Analytic Hierarchy Process (AHP) is as follows: Historical fault data for 3D printers includes the abnormal occurrence rate of various status parameters, the average over-limit amplitude of various status parameters, the average over-limit duration of various status parameters, the average fluctuation amplitude of various status parameters, the number of repairs for various faults, the maximum repair frequency for various faults, the average repair time for various faults, and the downtime impact duration of various faults. Obtain the anomaly occurrence rate weight factor, the average over-limit amplitude weight factor, the average over-limit duration weight factor, and the average parameter fluctuation amplitude weight factor from the database. Divide the anomaly occurrence rate, the average over-limit amplitude, the average over-limit duration, and the average parameter fluctuation amplitude of each type of state parameter by the corresponding preset threshold, then multiply by the corresponding weight factor, and finally add them together to obtain the anomaly index of each type of state parameter. The system retrieves the maintenance frequency weight factor, maximum maintenance frequency weight factor, average maintenance time weight factor, and downtime impact weight factor from the database. It then divides the maintenance frequency, maximum maintenance frequency, average maintenance time, and downtime impact of each type of fault by the corresponding preset threshold, multiplies them by the corresponding weight factor, and finally sums them to obtain the anomaly index of each type of fault. Based on the anomaly indices of various state parameters and various faults, correlation analysis is performed on various state parameters and various faults, and the maintenance weights of various state parameters are calculated.

4. The fault detection method in 3D printing according to claim 3, characterized in that, The correlation analysis is performed, and the specific analysis process is as follows: Using the fault index of a certain type of fault as the set of numerical content, and arranging them in the order of the statistical samples of historical faults, we obtain the fault index sequence of that type of fault, which is denoted as the baseline sequence of that type of fault. Based on the order of the statistical samples of historical faults of that type of fault, we obtain the abnormal index of a certain type of state parameter at each statistical time of historical faults of that type of fault, and arrange them in order to obtain the comparison sequence of that type of state parameter of that type of fault. The baseline sequence of this type of fault and the corresponding state parameter sequence to be compared are normalized to obtain the standard baseline sequence of this type of fault and the standard comparison sequence of the corresponding state parameter. The absolute difference between the baseline sequence value and the comparison sequence value under the same historical sample is calculated one by one, and the differences of each sample are integrated to form a difference set. The minimum and maximum differences are selected from the set of differences. Then, a preset fixed resolution coefficient is obtained from the database. The minimum difference is added to the product of the maximum difference and the fixed resolution coefficient to obtain the first-level correlation. The difference of each sample is added to the product of the maximum difference and the fixed resolution coefficient to obtain the first-level correlation of each sample. Then, the first-level correlation is divided by the first-level correlation of each sample to obtain the correlation coefficient of each sample. The arithmetic mean is calculated to obtain the final correlation. This gives the final correlation of the state parameters of this type of fault, and then the final correlation of the state parameters of various types of faults.

5. A fault detection method in 3D printing according to claim 4, characterized in that, The calculation yields the maintenance weights for various state parameters, and the specific calculation process is as follows: The various state parameters of each type of fault are sorted in descending order of final correlation degree to obtain the state parameter sequence of each type of fault. The first preset number of state parameters of each type of fault are recorded as the core correlation state parameters of each type of fault. The final correlation degree of each core correlation state parameter of each type of fault is added together to obtain the weight base of each type of fault. The final correlation degree of each type of fault state parameter is divided by the corresponding weight base to obtain the weight factor of each type of fault state parameter. The abnormal index of each state parameter of each type of fault is multiplied by the corresponding weight factor and then summed to obtain the detection fault index of each type of fault. The detection fault index and abnormal index of each type of fault are multiplied by the corresponding threshold, then multiplied by the corresponding weight factor, and finally summed to obtain the monitoring index of each type of fault. Based on the core associated status parameters of various faults, the core faults corresponding to each status parameter are obtained. The maximum monitoring index of each core fault corresponding to each status parameter is recorded as the monitoring index of each status parameter. The monitoring index of each status parameter is obtained by summing the monitoring indices of each status parameter. The maintenance weight of each status parameter is obtained by dividing the monitoring index of each status parameter by the total monitoring index.

6. A fault detection method in 3D printing according to claim 2, characterized in that, The entropy weight method, specifically the analysis process, is as follows: The values ​​of various status parameters corresponding to the occurrence of historical faults at a preset number of faults are summarized to obtain data sets of various status parameters. All data sets of status parameters are integrated into a total dataset to be calculated. Divide the total dataset by the corresponding threshold to obtain the normalized total dataset. Divide each individual data point under each parameter class by the sum of all data points under that parameter class to obtain the proportion of each data point. This gives the proportion of each data point for each type of state parameter. Substituting the proportions of each data point of various state parameters into the formula for calculating information entropy, we obtain the information entropy values ​​of various state parameters. The formula for calculating information entropy is: ,in Let be the information entropy value of the i-th type of state parameter, where i is the change in the state parameter, and ln is the natural logarithm. Let j represent the percentage of the j-th data point under the i-th state parameter category out of the total sum of all data points in that category, where j = 1, 2, ..., n, and n is a positive integer greater than 2. The total information entropy value is obtained by summing the information entropy values ​​of various state parameters. The objective weight of each state parameter is obtained by dividing the information entropy value of each state parameter by the total information entropy value. The weight factor of maintenance weight and the weight factor of objective weight are obtained from the database. The maintenance weight and objective weight of each state parameter are multiplied by the corresponding weight factor, and then summed to obtain the final weight value of each state parameter.

7. A fault detection method in 3D printing according to claim 2, characterized in that, The specific implementation process of the 3D printer status acquisition scheme is as follows: Staff preset the basic acquisition cycle for various status parameters, and sort the status parameters in descending order of their final weight values ​​to obtain a status parameter acquisition sequence. The first preset number of various status parameters in the status parameter acquisition sequence are marked as key status parameters. Staff preset the acquisition cycle correction factor for each position in the sequence, and multiply the basic acquisition cycle for each status parameter by the corresponding acquisition cycle correction factor to obtain the corrected acquisition cycle for each status parameter. The 3D printer status acquisition scheme is as follows: data for each status parameter is acquired according to the corresponding corrected acquisition cycle.

8. A fault detection method in 3D printing according to claim 2, characterized in that, The specific implementation process of the 3D printer status analysis scheme is as follows: Based on the final weight values ​​of various state parameters, when abnormal state parameters occur, various abnormal state parameters are weighted and analyzed according to their corresponding final weight values.

9. A fault detection method in 3D printing according to claim 1, characterized in that, The analysis of the current 3D printer status data is as follows: The current 3D printer status data includes various status data collected each time within the current preset time period; According to the 3D printer status acquisition scheme, various types of status data are acquired in each acquisition within the current preset time period. If a certain type of status data acquired in a certain acquisition within the current preset time period is greater than the corresponding preset first threshold, the status data acquired in that acquisition within the current preset time period is recorded as abnormal data, and the current types of abnormal data are obtained. The average value of each type of state data collected within the current preset time period is obtained by averaging the data. If the average value of a certain type of state data within the current preset time period is greater than the corresponding preset second threshold, it is recorded as periodic abnormal data. In this way, various types of periodic abnormal data and various types of detection abnormal data within the current preset time period are obtained.

10. A fault detection method in 3D printing according to claim 9, characterized in that, The status warning for the 3D printer is implemented, and the specific warning process is as follows: The final weight values ​​corresponding to various types of abnormal data within the current preset time period are added together to obtain the current total final weight value. The value of each type of abnormal data is subtracted from the corresponding threshold, and then divided by the corresponding threshold. After absolute value conversion, the abnormality rate of each type of abnormal data is obtained. The abnormality rate of each type of abnormal data is multiplied by the corresponding final weight value, and then divided by the current total final weight value to obtain the current fault detection index. If the current fault detection index is greater than the preset current fault detection index threshold, an early warning is issued. At the same time, if a certain type of periodic abnormal data appears within the current preset time period, an early warning will be issued directly.

Images