Spcc method for multi-model product production process

Abstract

The SPC method for multi-model product production process relates to the technical field of production process monitoring in manufacturing industry, and judges whether the nominal values of various models are same, whether the upper tolerances are same, and whether the lower tolerances are same, processes the detection data of various model products, so that the processed detection data is in the same normal distribution, thereby drawing all the processed detection data of all model products in the same SPC control chart, and using the differentiating rule to differentiate the stability of the production process according to the SPC control chart, so as to monitor the stability of the production process, solve the problem that the differentiating rule cannot be used to monitor the stability of the multi-model product production process, and the present application is suitable for monitoring the stability of the multi-model product production process.

Description

Technical Field

[0001] This invention relates to the field of production process monitoring technology in the manufacturing industry, and particularly to the SPC method for the production process of multiple product models. Background Technology

[0002] SPC, or Statistical Process Control, is a process control tool that uses mathematical statistics to provide feedback on quality fluctuations in the production process, identify potential quality problems based on anomaly detection rules, conduct root cause analysis, and formulate corresponding control measures to eliminate them. This allows for the assessment and monitoring of the stability of the production process.

[0003] The existing approach involves using certain data transformation methods to convert the raw data into a specific uniform specification, keeping the data within a small range, such as 0 to 1 or -1 to 1, to eliminate differences in the properties, dimensions, orders of magnitude, and other attributes of different variables. This transforms the data into dimensionless standardized values, ensuring that the values ​​of each indicator are on the same order of magnitude, which facilitates comprehensive analysis and comparison between indicators. However, there is a lack of control chart monitoring methods for situations where there are multiple models and different specifications in the same production process and the sample size is greater than 1.

[0004] Manufacturing enterprises require the same production line to be able to switch between producing different specifications and models of products. According to quality management requirements, SPC monitoring and analysis of the production process is necessary. Since the upper and lower limits of the specifications of each model are not uniform, the upper and lower control limits will also be inconsistent. Therefore, if all the test data of all models are plotted on the same SPC control chart, there will be a large span, which makes it impossible to use the anomaly detection rules to detect anomalies and monitor the stability of the production process. Summary of the Invention

[0005] The technical problem solved by this invention is to provide an SPC method for the production process of multiple product models, which solves the problem that the stability of the production process of multiple product models cannot be monitored using outlier detection rules.

[0006] The technical solution adopted by this invention to solve the above-mentioned technical problems is an SPC method for the production process of multiple product models, comprising the following steps:

[0007] S1. Obtain the specifications for each product model, including nominal values, upper tolerances, and lower tolerances;

[0008] S2. Collect test data for each model of product according to the preset frequency, and determine whether the nominal values, upper tolerances, and lower tolerances of each model of product are the same.

[0009] If the nominal values, upper tolerances, and lower tolerances are the same, then the test data of all models of products should be plotted on the same SPC control chart.

[0010] If the nominal values ​​are different, the upper tolerance is the same, and the lower tolerance is the same, then the nominal values ​​are used to process each test data to obtain all the processed test data, and all the processed test data of all models of products are plotted on the same SPC control chart.

[0011] If at least one of the upper and lower tolerances is different, the average and standard deviation estimates of all test data of the selected sample are used to process each test data of each model of product to obtain all the processed test data, and all the processed test data of all models of products are plotted on the same SPC control chart.

[0012] S3. Based on the SPC control chart, use the anomaly detection rules to detect anomalies in the production process and monitor its stability.

[0013] Furthermore, the formula for processing each test data using the nominal value is: Y = XD, where Y represents the processed test data of a certain model product, X represents the test data of a certain model product, and D represents the nominal value of the certain model product.

[0014] Furthermore, the formula for processing each test data for each model of product using the estimated mean and standard deviation of all test data from the selected samples is as follows: Where Y represents a certain test data of a certain model product after processing, X represents a certain test data of a certain model product, E represents the average value of all test data of the selected sample of the certain model product, and F represents the estimated standard deviation of all test data of the selected sample of the certain model product.

[0015] Furthermore, the estimated standard deviation is equal to the mean of the ranges divided by a constant d2, which is obtained by querying the factor table of the control limits of the econometric control chart.

[0016] Furthermore, the estimated standard deviation is equal to the mean standard deviation divided by a constant C4, which is obtained by querying the factor table of the control limits of the measurement control chart.

[0017] The beneficial effects of this invention are as follows: The SPC method for the production process of multiple product models in this invention processes the test data of each product model by judging whether the nominal values, upper tolerances, and lower tolerances of each model are the same, so that the processed test data are in the same normal distribution. Thus, all the processed test data of all product models are plotted on the same SPC control chart. Based on the SPC control chart, the stability of the production process is judged and monitored using outlier rules, which solves the problem that the stability of the production process of multiple product models cannot be monitored using outlier rules. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the SPC method for manufacturing multiple product models according to the present invention.

[0019] Figure 2 This is an example of plotting test data from multiple product models onto the same SPC control chart.

[0020] Figure 3 This is an SPC control chart obtained using the SPC method for the production process of multiple product models according to the present invention. Detailed Implementation

[0021] The present invention provides an SPC method for the production process of multiple product models, such as... Figure 1 As shown, it includes the following steps:

[0022] S1. Obtain the specifications for each product model, including nominal values, upper tolerances, and lower tolerances;

[0023] Specifically, the specifications are as follows: The nominal value is 100, the upper tolerance is 0.05, and the lower tolerance is 0.05.

[0024] S2. Collect test data for each model of product according to the preset frequency, and determine whether the nominal values, upper tolerances, and lower tolerances of each model of product are the same.

[0025] If the nominal values, upper tolerances, and lower tolerances are the same, then the test data of all models of products should be plotted on the same SPC control chart.

[0026] If the nominal values ​​are different, the upper tolerance is the same, and the lower tolerance is the same, then the nominal values ​​are used to process each test data to obtain all processed test data. All processed test data for all models of products are plotted on the same SPC control chart. Specifically, the formula for processing each test data using the nominal values ​​is: Y = XD, where Y represents the processed test data for a certain model of product, X represents the test data for a certain model of product, and D represents the nominal value of the certain model of product.

[0027] If at least one of the upper and lower tolerances is different, then the average and standard deviation estimates of all test data from the selected sample are used to process each test data for each product model to obtain all processed test data. All processed test data for all product models are then plotted on the same SPC control chart. Specifically, the formula for processing each test data for each product model using the average and standard deviation estimates of all test data from the selected sample is as follows: Where Y represents a certain test data of a certain model product after processing, X represents a certain test data of a certain model product, E represents the average value of all test data of the selected samples of the certain model product, and F represents the estimated standard deviation of all test data of the selected samples of the certain model product. The estimated standard deviation is equal to the mean of the range values ​​divided by a constant d2. The constant d2 is obtained by querying the factor table of the control limits of the metrological control chart. Alternatively, the estimated standard deviation is equal to the mean of the standard deviation divided by a constant C4. The constant C4 is obtained by querying the factor table of the control limits of the metrological control chart, which is derived from GB / T 17989.2-2020.

[0028] S3. Based on the SPC control chart, use the anomaly detection rules to detect anomalies in the production process and monitor its stability.

[0029] To further illustrate the invention, consider an example: a crankshaft assembly for grinding the roundness of a long shaft in a machining workshop. Three models are designated A, B, and C, with the following specifications:

[0030] Model A: Upper specification value 0.015, lower specification value 0.000, nominal value 0.0075;

[0031] Model B: Upper specification value 0.020, lower specification value 0.000, nominal value 0.010;

[0032] Model C: Upper specification value 0.030, lower specification value 0.000, nominal value 0.015;

[0033] Calculate the tolerances for the three models:

[0034] Tolerance values ​​for model A: Upper tolerance = 0.015 - 0.0075 = 0.0075, Lower tolerance = 0.0075 - 0.000 = 0.0075;

[0035] Tolerance values ​​for model B: Upper tolerance = 0.020 - 0.010 = 0.010, Lower tolerance = 0.010 - 0.000 = 0.010;

[0036] Type C tolerance values: Upper tolerance = 0.030 - 0.015 = 0.015, Lower tolerance = 0.015 - 0.000 = 0.015;

[0037] Test data for each product model was collected according to a preset frequency. Three samples were selected for each model, and a total of 10 test data were collected, denoted as serial numbers 1-10. Test data for model A:

[0038] Sample 1 0.009 0.0135 0.0096 0.021 0.0071 0.012 0.0136 0.009 0.0126 0.01 Sample 2 0.01 0.01 0.0111 0.0113 0.0098 0.0094 0.01 0.0111 0.01 0.0119 Sample 3 0.0083 0.0103 0.013 0.0122 0.0126 0.0096 0.0147 0.01 0.0101 0.0104

[0039] Test data for Model B:

[0040] Sample 1 0.0117 0.0162 0.0123 0.0237 0.0098 0.0147 0.0163 0.0117 0.0153 0.0127 Sample 2 0.0126 0.0126 0.0137 0.0139 0.0124 0.012 0.0126 0.0137 0.0126 0.0145 Sample 3 0.0111 0.0131 0.0158 0.015 0.0154 0.0124 0.0175 0.0128 0.0129 0.0132

[0041] Test data for Model C:

[0042] Sample 1 0.0189 0.0235 0.0195 0.0309 0.0171 0.0216 0.0235 0.019 0.0225 0.02 Sample 2 0.0199 0.0199 0.0208 0.0212 0.0197 0.0194 0.0199 0.021 0.0198 0.0218 Sample 3 0.0182 0.0203 0.0229 0.0221 0.0226 0.0195 0.0246 0.0199 0.02 0.0204

[0043] If the test data for models A, B, and C are not processed, and the average and range values ​​of each group of test data for different models are calculated and plotted on the same SPC control chart, the result will be as follows: Figure 2 The SPC mean control chart shown demonstrates that plotting these values ​​on the same control chart can result in large variations, making it impossible to monitor the stability of the production process for multiple product models using outlier detection rules.

[0044] If the SPC method for manufacturing multiple product models provided by this invention is used, then it is determined whether the nominal values, upper tolerances, and lower tolerances of each product model are the same. Since the nominal value of model A ≠ the nominal value of model B ≠ the nominal value of model C; and the tolerance value of model A ≠ the tolerance value of model B ≠ the tolerance value of model C, therefore, the formula is used... Each test data for each product model is processed, where Y represents a processed test data for a certain product model, X represents a test data for a certain product model, E represents the average value of all test data for the selected sample of the product model, and F represents the estimated standard deviation of all test data for the selected sample of the product model.

[0045] The following values ​​were calculated from the test data collected from the three samples: For model A, the average value of all test data was 0.01111, the mean range was 0.00377, and the estimated standard deviation was 0.00223; for model B, the average value was 0.01381, the mean range was 0.00381, and the estimated standard deviation was 0.00225; for model C, the average value was 0.02101, the mean range was 0.00373, and the estimated standard deviation was 0.0022. The test data were processed. For example, the first test data of sample 1 for model A, which was 0.009, was processed to (0.009-0.01111) / 0.00223 = -0.94619. This process was repeated for each test data of each model, resulting in the following processed test data:

[0046] Test data after processing (Type A):

[0047]

[0048] Test data after processing (Type B):

[0049]

[0050] Test data after C-type processing:

[0051]

[0052] All test data after processing of all product models are plotted on the same SPC control chart, such as... Figure 3 The SPC mean control chart shown shows that the data obtained in the SPC mean control chart are all in a normal distribution. The outlier detection rules can be used to detect outliers in the production process and monitor the stability of the production process.

Claims

1. An SPC method for manufacturing multiple product models, characterized in that, Includes the following steps: S1. Obtain the specifications for each product model, including nominal values, upper tolerances, and lower tolerances; S2. Collect test data for each model of product according to the preset frequency, and determine whether the nominal values, upper tolerances, and lower tolerances of each model of product are the same. If the nominal values, upper tolerances, and lower tolerances are the same, then the test data of all models of products should be plotted on the same SPC control chart. If the nominal values ​​are different, the upper tolerance is the same, and the lower tolerance is the same, then the nominal values ​​are used to process each test data to obtain all the processed test data, and all the processed test data of all models of products are plotted on the same SPC control chart. If at least one of the upper and lower tolerances is different, the average and standard deviation estimates of all test data of the selected sample are used to process each test data of each model of product to obtain all the processed test data, and all the processed test data of all models of products are plotted on the same SPC control chart. S3. Based on the SPC control chart, use the anomaly detection rules to detect anomalies in the production process and monitor its stability.

2. The SPC method for manufacturing multiple product models according to claim 1, characterized in that, The formula for processing each test data using the nominal value is: Y = XD, where Y represents the processed test data of a certain model of product, X represents the test data of a certain model of product, and D represents the nominal value of the certain model of product.

3. The SPC method for manufacturing multiple product models according to claim 1, characterized in that, The formula for processing each test data point for each product model using the estimated mean and standard deviation of all test data from the selected sample is as follows: Where Y represents a certain test data of a certain model product after processing, X represents a certain test data of a certain model product, E represents the average value of all test data of the selected sample of the certain model product, and F represents the estimated standard deviation of all test data of the selected sample of the certain model product.

4. The SPC method for manufacturing multiple product models according to claim 3, characterized in that, The estimated standard deviation is equal to the mean of the ranges divided by a constant d2, which is obtained by querying the factor table of the control limits of the econometric control chart.

5. The SPC method for manufacturing multiple product models according to claim 3, characterized in that, The estimated standard deviation is equal to the mean standard deviation divided by a constant C4, which is obtained by querying the factor table of the control limits of the econometric control chart.

Images