Engine finished product part automatic detection device and method
Through the combination of ID reading and robot trajectory motion control, automatic detection of engine finished product components is realized, and the visual fatigue and multi-model identification problems of visual inspection in the prior art are solved, thereby improving detection efficiency and accuracy.
Patent Information
- Application Number
- CN202410843102.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-06-27
- Publication Date
- 2024-11-08
AI Technical Summary
In the prior art, visual inspection of finished product parts of the engine requires long-term scanning and movement of the operator, resulting in increased visual fatigue and operation intensity, and it is difficult to effectively determine whether the components of various vehicle models are properly installed.
The engine model is distinguished by ID reading method, and the fixed camera is transformed into a movable camera through the built-in track motion control of the robot. The logical program is used to determine the results of multiple sets of photography to determine whether the engine finished parts are from the database. The stored qualified products are consistent, and the judgment results are output to the screen through communication and saved.
It reduces the operator's visual work intensity, improves the inspection quality, eliminates the outflow of bad products, and is suitable for the detection of multi-component engines and the switching and judgment of multiple models.
Smart Images

Figure CN118918346A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of engine component detection, and in particular relates to an automatic detection device and method for engine finished product components. Background Art
[0002] The TNGA #5 assembly line started production in July 2021 and went into mass production in November, producing 2.0L and 2.5L engines for a total of 7 models. In order to ensure the quality of the finished product, each engine requires operators to conduct visual inspections to determine whether the parts are installed incorrectly or not installed. There are many models, and the operators confirm and learn the parts of each model.
[0003] Through the above analysis, the problems and defects of the existing technology are as follows: the operator needs to visually check and use a paint pen to increase the work intensity. The operator has to move a long scanning distance for visual confirmation, which can easily lead to visual fatigue and misjudgment. Summary of the invention
[0004] In order to overcome the problems existing in the related art, the disclosed embodiments of the present invention provide a device and method for automatically detecting engine finished product parts, specifically for performing component quality determination and detection on 7 types of engines of TNGA2.0 / 2.5.
[0005] The technical solution is as follows: A method for automatically detecting engine finished product parts, comprising:
[0006] S1, using the ID reading method to identify the engine model, and sending a camera switching instruction to the camera scene group switching program;
[0007] S2, using the robot's built-in trajectory motion control method to transform the fixed camera into a movable camera, and using the robot's movement to reach the specified position;
[0008] S3, judging the multiple groups of photographic results through a logic program to determine whether the engine finished product parts are consistent with the qualified products stored in the database;
[0009] S4, output the determination result to the screen through communication means, and save the detection data.
[0010] In step S1, the engine model is identified by using an ID reading method, including:
[0011] In the first step, the ID reading matrix and the engine coding matrix of 7 types of models, including 2.0L and 2.5L, are used to encode the transmitted captured image signals of the camera scene group;
[0012] In the second step, an optimal ratio combined image resolver is used to decode the transmitted captured image signal sent by the camera scene group.
[0013] Furthermore, in the first step, encoding the transmitted captured image signal of the camera scene group includes:
[0014] The regional single resolution consists of the ID reading matrix Q and the engine coding matrix C of 7 types of models including 2.0L and 2.5L. The expression is:
[0015] Q=I M ×Q0
[0016] C=I M ×C0
[0017] Where Q is the ID reading matrix, C is the engine coding matrix, I M is the resolution of the engine image of different models, I M =(1, 2…M) T , M is the different code symbols of the engines of 7 types of vehicles including 2.OL, 2.5L, T is the number of times the camera scene group is photographed; Q0 is the unitary matrix, and C0 is the orthogonal space-time code.
[0018] Furthermore, when the unitary matrix Q0 and the orthogonal space-time code C0 select a 7×7 model, we have:
[0019]
[0020] The ID reading matrix Q of the camera scene group and the encoding matrix C of the tag end are:
[0021]
[0022] In the formula, c1 is the engine code value of the first type of vehicle, c7 is the engine code value of the seventh type of vehicle, is the encoding value mapped in the orthogonal of the engine of the first model, It is the coding value mapped in the orthogonal of the engine of the 7th model.
[0023] In the second step, the transmitted captured image signal of the camera scene group is encoded, and the expression is:
[0024] R=(I M ×Q0)×(I M ×C0)H T +W
[0025] In the formula, R is the captured image signal code, H T is the orthogonal space-time code, and W is the adjustment coefficient.
[0026] Furthermore, the engine codes and image resolutions of 2.0L and 2.5L models are derived by analogy.
[0027] In step S2, the fixed camera is changed into a movable camera using the trajectory motion control method built into the robot, and the robot is moved to reach the designated position, including:
[0028] Step 1, obtaining a certain type or several types of trajectory motion vector mode camera trajectory change graphics, using a standard graphics library to convert the format, obtaining the corresponding image mapping format camera trajectory change graphics, and constructing a new data set using the trajectory motion vector mode camera trajectory change graphics and the image mapping format camera trajectory change graphics;
[0029] Step 2: Use the convolutional neural network and attention module to obtain the global and local feature structures of the trajectory vectors in the new dataset;
[0030] Step 3, the global and local feature structures of the trajectory route vector and the trajectory motion vector pattern camera trajectory change graph are input into a decoder composed of a long short-term memory network to obtain the trajectory route normal distribution parameter data and trajectory route state classification distribution probability data of the new data set;
[0031] Step 4, randomly sampling from normal distribution to obtain the trajectory route position data of each scanning moving distance step, calculating the trajectory route state data of each scanning moving distance step from the trajectory route state classification step probability data, combining the output of each scanning moving distance step, and obtaining the combined predicted camera trajectory change graph;
[0032] Step 5, inputting the generated camera trajectory change graph and the trajectory motion vector mode camera trajectory change graph into a discriminator composed of a long short-term memory network, obtaining deviation information of the discriminator predicting the input data, and obtaining a camera trajectory change graph generation model integrating a convolutional neural network and a long short-term memory network after training is completed;
[0033] Step six, input the camera trajectory change graph of the trajectory motion vector pattern to be predicted and the corresponding image mapping format camera trajectory change graph into the camera trajectory change graph generation model integrating the convolutional neural network and the long short-term memory network for prediction, and obtain the final robot prediction generation camera trajectory change graph.
[0034] In step S3, determining whether the finished engine component is consistent with the qualified product stored in the database includes:
[0035] S301, obtaining response determination curve data of multiple groups of cameras for recognition of engine finished product parts;
[0036] S302, extracting characteristic values of engine finished product components from the response determination curve;
[0037] S303, preprocessing and standardizing the characteristic values to obtain a characteristic data set;
[0038] S304, randomly dividing the feature data set into a training data set and a test data set;
[0039] S305, using a deep learning network to train the training data set to obtain a comprehensive engine finished product component classification model, and using a test data set to test the comprehensive engine finished product component classification model, based on the test results and the qualified products stored in the database are compared to determine whether they are consistent, the classification types include 7 types of engines for 2.0L and 2.5L models.
[0040] Another object of the present invention is to provide an automatic detection device for engine finished product parts, which is implemented by the automatic detection method for engine finished product parts, and the device comprises:
[0041] The camera scene group is used to identify the engine model by using the ID reading method and send a camera switching instruction to the camera scene group switching program;
[0042] Robot, used for built-in trajectory motion control method to transform the fixed camera into a movable camera, using the movement of the robot to reach the specified position;
[0043] A logic program judgment module is used to judge multiple groups of photographic results through logic programs to determine whether the engine finished product parts are consistent with qualified products stored in the database;
[0044] The communication display module is used to output the judgment results to the screen through communication; and to save the detection data.
[0045] Furthermore, a light-proof dark brown glass protective plate is installed on the outside of the automatic inspection device for finished engine parts.
[0046] Combining all the above technical solutions, the beneficial effects of the present invention are as follows: the present invention uses the mode of distinguishing between vehicle models, the movability of photography, the display output after the judgment of the logic program, and the dark brown glass shield for light protection; and uses the program to distinguish engines of different models, uses the movement of the robot to reach the specified position, and judges the results of the camera through the logic program and displays them on the screen; it reduces the visual work intensity of the operator. Improve the detection quality and prevent the outflow of defective products; it is suitable for the detection of multi-component engines. It is suitable for multi-model switching and discrimination detection. BRIEF DESCRIPTION OF THE DRAWINGS
[0047] The accompanying drawings herein are incorporated in and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the description, serve to explain the principles of the present disclosure;
[0048] Figure 1 is a flow chart of an automatic detection method for engine finished product parts provided by an embodiment of the present invention;
[0049] Figure 2 It is a schematic diagram of the automatic detection method of engine finished product parts provided by an embodiment of the present invention;
[0050] Figure 3 Schematic diagram of an automatic detection device for engine finished product parts provided by an embodiment of the present invention;
[0051] Figure 4 This is a diagram of an application scenario of the automatic detection device for engine finished product parts provided by an embodiment of the present invention;
[0052] In the figure: 1. Camera scene group; 2. Robot; 3. Logic program judgment module; 4. Communication display module. DETAILED DESCRIPTION
[0053] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without violating the connotation of the present invention, so the present invention is not limited by the specific implementation disclosed below.
[0054] Embodiment 1, as Figure 1 As shown, the automatic detection method for engine finished product parts includes:
[0055] S1, using the ID reading method to identify the engine model, and sending a camera switching instruction to the camera scene group switching program;
[0056] S2, using the robot's built-in trajectory motion control method to transform the fixed camera into a movable camera, and using the robot's movement to reach the specified position;
[0057] S3, judging the multiple groups of photographic results through a logic program to determine whether the engine finished product parts are consistent with the qualified products stored in the database;
[0058] S4, output the determination result to the screen through communication means, and save the detection data.
[0059] Embodiment 2, as a preferred embodiment, in step S1, distinguishing the engine model by using the ID reading method includes:
[0060] In the first step, the ID reading matrix and the engine coding matrix of 7 types of models, including 2.0L and 2.5L, are used to encode the transmitted captured image signals of the camera scene group;
[0061] In the second step, an optimal ratio combined image resolver is used to decode the transmitted captured image signal sent by the camera scene group.
[0062] In the first step, the transmitted captured image signal of the camera scene group is encoded, including:
[0063] The regional single resolution consists of the ID reading matrix Q and the engine coding matrix C of 7 types of models including 2.0L and 2.5L. The expression is:
[0064] Q=I M ×Q0
[0065] C=I M ×C0
[0066] Where Q is the ID reading matrix, C is the engine coding matrix, I M is the resolution of the engine image of different models, I M =(1, 2…M) T , M is the different code symbols of the engines of 7 types of 2.0L and 2.5L models, T is the number of times the camera scene group is photographed; Q0 is the unitary matrix, and C0 is the orthogonal space-time code.
[0067] When the unitary matrix Q0 and the orthogonal space-time code C0 select a 7×7 model, we have:
[0068]
[0069] The ID reading matrix Q of the camera scene group and the encoding matrix C of the tag end are:
[0070]
[0071] In the formula, c1 is the engine code value of the first type of vehicle, c7 is the engine code value of the seventh type of vehicle, is the encoding value mapped in the orthogonal of the engine of the first model, It is the coding value mapped in the orthogonal of the engine of the 7th model.
[0072] In the second step, the transmitted captured image signal of the camera scene group is encoded, and the expression is:
[0073] R=(I M ×Q0)×(I M ×C0)H T +W
[0074] Where R is the captured image signal code, R T is the orthogonal space-time code, and W is the adjustment coefficient.
[0075] The engine codes and image resolutions of other 2.0L and 2.5L models are similar.
[0076] Embodiment 3, as a preferred embodiment, in step S2, the robot built-in trajectory motion control method changes the fixed camera into a movable camera, and the trajectory motion control method includes:
[0077] Step 1, obtaining a certain type or several types of trajectory motion vector mode camera trajectory change graphics, using a standard graphics library to convert the format, obtaining the corresponding image mapping format camera trajectory change graphics, and constructing a new data set using the trajectory motion vector mode camera trajectory change graphics and the image mapping format camera trajectory change graphics;
[0078] Step 2: Use the convolutional neural network and attention module to obtain the global and local feature structures of the trajectory vectors in the new dataset;
[0079] Step 3, the global and local feature structures of the trajectory route vector and the trajectory motion vector pattern camera trajectory change graph are input into a decoder composed of a long short-term memory network to obtain the trajectory route normal distribution parameter data and trajectory route state classification distribution probability data of the new data set;
[0080] Step 4, randomly sampling from normal distribution to obtain the trajectory route position data of each scanning moving distance step, calculating the trajectory route state data of each scanning moving distance step from the trajectory route state classification step probability data, combining the output of each scanning moving distance step, and obtaining the combined predicted camera trajectory change graph;
[0081] Step 5, inputting the generated camera trajectory change graph and the trajectory motion vector mode camera trajectory change graph into a discriminator composed of a long short-term memory network, obtaining deviation information of the discriminator predicting the input data, and obtaining a camera trajectory change graph generation model integrating a convolutional neural network and a long short-term memory network after training is completed;
[0082] Step six, input the camera trajectory change graph of the trajectory motion vector pattern to be predicted and the corresponding image mapping format camera trajectory change graph into the camera trajectory change graph generation model integrating the convolutional neural network and the long short-term memory network for prediction, and obtain the final robot prediction generation camera trajectory change graph.
[0083] In the step 1, the trajectory motion vector mode camera trajectory change graph is composed of a triple consisting of a data point offset and trajectory route state information;
[0084] In the step 1, the method of constructing a new data set using the camera trajectory change graph in the trajectory motion vector mode and the camera trajectory change graph in the image mapping format includes:
[0085] (1) obtaining a camera trajectory change graph of a trajectory motion vector pattern in the form of a triple expression, and converting the camera trajectory change graph of the trajectory motion vector pattern into an image mapping format camera trajectory change graph using a format conversion library;
[0086] (2) converting the camera trajectory change graph of the trajectory motion vector pattern in the triple expression form into a triple expression form, and obtaining the maximum component length of the camera trajectory change graph in the data set;
[0087] In step 2, the global feature structure of the trajectory route vector in the new data set obtained by using the convolutional neural network and the attention module includes:
[0088] 1) Inputting the obtained image mapping format camera trajectory change graph into the convolutional neural network for training, performing linear calculation on the output of the convolutional neural network to obtain the parameters of the global and local feature structures constituting the trajectory route vector, performing linear calculation and sampling on the parameters to obtain the local position feature data of the trajectory route vector;
[0089] 2) Add an attention module after the first layer of the convolutional neural network to obtain the global and local feature structure data of the trajectory route vector;
[0090] In step 2), the convolutional neural network includes a neural network layer for obtaining a global feature structure, and the convolutional neural network has a total of 7 layers;
[0091] In step 2), the training is performed using dropout, and the dropout ratio is 0.9;
[0092] In step 3, the method for obtaining the normal distribution parameter data of the trajectory route includes:
[0093] In the first step, the obtained global and local feature structure data are input into the decoder built by LSTM together with the vector camera trajectory change graph as noise to obtain the output of each scanning movement distance step;
[0094] Step 2: Input the output of each scanning moving distance step into the Gaussian mixture model to obtain the parameters constituting the bivariate normal distribution and model the trajectory route state data;
[0095] Step 3, the obtained parameters constituting the bivariate normal distribution are used to obtain the bivariate normal distribution function, and the trajectory route position data is obtained by sampling, and the trajectory route state data of the current scanning moving distance step is calculated from the trajectory route state classification distribution probability model;
[0096] Step 4: The obtained trajectory position data and state data are combined into a predicted output trajectory in the form of a triple expression; the predicted output trajectory of all scanning moving distance steps are combined to obtain a predicted output camera trajectory change graph.
[0097] The invention has good feature extraction effect, high quality of trajectory map and more practical generated trajectory route.
[0098] Embodiment 4, as a preferred embodiment, in step S3, judging multiple groups of photographic results by a logic program includes: Further, the method for intelligent olfactory recognition of engine finished product parts includes the following steps:
[0099] S301, obtaining response determination curve data of multiple groups of cameras for recognition of engine finished product parts;
[0100] S302, extracting characteristic values of engine finished product components from the response determination curve;
[0101] S303, preprocessing and standardizing the characteristic values to obtain a characteristic data set;
[0102] S304, randomly dividing the feature data set into a training data set and a test data set;
[0103] S305, using a deep learning network to train the training data set to obtain a comprehensive engine finished product component classification model, and using a test data set to test the comprehensive engine finished product component classification model, based on the test results and the qualified products stored in the database are compared to determine whether they are consistent, the classification types include 7 types of engines for 2.0L and 2.5L models.
[0104] The characteristic values in step S302 include response smoothness, determination smoothness, steady-state cleanliness, the sum of curve integral crack gaps, maximum smoothness value, maximum positive slope in the rising stage, and minimum negative slope in the falling stage;
[0105] The response smoothness is the smoothness required when the output change reaches a stable value of 0.01% when the quality characteristic value of the finished engine component scanned by the camera group undergoes a step change; the judgment smoothness is the smoothness required for the camera group to judge the equilibrium state to 90% of the signal; the steady-state cleanliness is the cleanliness of the camera group response curve to maintain the equilibrium state.
[0106] The sum of the curve-integrated crack gaps is calculated using a quadrature formula;
[0107]
[0108] In the formula, the sum of the S-curve integral crack gap, f(x i ), f(x i+1 ) are the i-th and i+1-th function values on the curve, n is the number of parts of the integration area, f(x) is the integration step, a and b are the positions between the horizontal coordinates of two adjacent points on the curve.
[0109] The formula for calculating the maximum positive slope in the rising phase is as follows:
[0110]
[0111] In the formula, k max is the maximum smoothness value, and Δx is the rising stage area.
[0112] In step S303, preprocessing and standardizing the characteristic values extracted from the response determination curve includes:
[0113] According to the preset useless features, remove useless data to obtain the feature value;
[0114] By using standardization, the eigenvalues are dimensionless.
[0115] The deep learning network in step 305 includes deep learning networks including logistic regression and artificial neural networks.
[0116] The present invention uses a deep learning network to train feature data sets collected by multiple groups of cameras to obtain a comprehensive classification model, and then obtains the prediction results of the types of engine finished product parts. The comprehensive engine finished product component classification model provided by the present invention uses a logistic regression artificial neural network algorithm; accordingly, the comprehensive engine finished product component classification model will detect the engine finished product components to be tested and obtain an optimal result. Through simulation experiments, it can be seen that the discrimination accuracy provided by the present invention has reached more than 93%, among which the artificial neural network has the best effect, with an accuracy rate of 96.5%.
[0117] The present invention trains the feature data of engine finished product parts, and predicts and classifies the engine finished product parts data to be tested through a comprehensive engine finished product parts classification model, thereby reducing human subjective participation, thereby being able to accurately identify the category and quality status of the engine finished product parts.
[0118] Embodiment 5, Figure 2 The principle of the method for automatically detecting engine finished product parts provided by the embodiment of the present invention includes:
[0119] Arrival signal is issued;
[0120] ID reading;
[0121] Camera scene group switching; robot program selection at the same time;
[0122] The robot moves into position;
[0123] Camera scene switching;
[0124] Take photos;
[0125] The judgment is OK or NG, and the screen displays OK or NG.
[0126] Embodiment 6, as Figure 3 As shown, an embodiment of the present invention provides an automatic detection device for engine finished product parts, comprising:
[0127] Camera scene group 1 is used to identify the engine model using the ID reading method and send a camera switching instruction to the camera scene group switching program;
[0128] Robot 2, used for the built-in trajectory motion control method to transform the fixed camera into a movable camera, using the movement of the robot to reach the specified position;
[0129] The logic program judgment module 3 is used to judge the multiple groups of photographic results through the logic program to determine whether the engine finished product parts are consistent with the qualified products stored in the database;
[0130] The communication display module 4 is used to output the determination result to the screen through communication means; and to save the detection data.
[0131] In order to ensure the detection effect, a light-proof dark brown glass protective plate can be installed on the outside of the automatic detection device for finished engine parts.
[0132] in, Figure 4 This is a diagram of an application scenario of the automatic detection device for finished engine parts provided in an embodiment of the present invention.
[0133] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.
[0134] Since the information interaction, execution process, etc. between the above-mentioned devices / units are based on the same concept as the method embodiment of the present invention, their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.
[0135] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated in one processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment.
[0136] An embodiment of the present invention further provides a computer device, which includes: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the processor implements the steps in any of the above-mentioned method embodiments when executing the computer program.
[0137] An embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the above-mentioned method embodiments can be implemented.
[0138] An embodiment of the present invention further provides an information data processing terminal, which, when executed on an electronic device, provides a user input interface to implement the steps in the above method embodiments. The information data processing terminal is not limited to mobile phones, computers, and switches.
[0139] An embodiment of the present invention further provides a server, which is used to provide a user input interface to implement the steps in the above method embodiments when executed on an electronic device.
[0140] An embodiment of the present invention provides a computer program product. When the computer program product runs on an electronic device, the electronic device can implement the steps in the above-mentioned method embodiments when executing the computer program product.
[0141] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, which can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may at least include: any entity or device that can carry the computer program code to the camera / terminal device, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. For example, a USB flash drive, a mobile hard disk, a disk or an optical disk.
[0142] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by any technician familiar with the technical field within the technical scope disclosed by the present invention and within the spirit and principles of the present invention should be covered within the protection scope of the present invention.
Claims
1. A method for automatically detecting engine finished parts, characterized in that: The method includes: S1, using the ID reading method to identify the engine model, and sending a camera switching instruction to the camera scene group switching program; S2, using the robot's built-in trajectory motion control method to transform the fixed camera into a movable camera, and using the robot's movement to reach the specified position; S3, judging the multiple groups of photographic results through a logic program to determine whether the engine finished product parts are consistent with the qualified products stored in the database; S4, output the determination result to the screen through communication means, and save the detection data.
2. The method for automatically inspecting engine finished parts according to claim 1, characterized in that: In step S1, the engine model is identified by using an ID reading method, including: In the first step, the ID reading matrix and the engine coding matrix of 7 types of models, including 2.0L and 2.5L, are used to encode the transmitted captured image signals of the camera scene group; In the second step, an optimal ratio combined image resolver is used to decode the transmitted captured image signal sent by the camera scene group.
3. The method for automatically inspecting engine finished parts according to claim 2, characterized in that: In the first step, the captured image signal transmitted by the camera scene group is encoded, including: The regional single resolution consists of the ID reading matrix Q and the engine coding matrix C of 7 types of models including 2.0L and 2.5L. The expression is: Q=I M ×Q0 C=I M ×C0 Where Q is the ID reading matrix, C is the engine coding matrix, I M is the resolution of the engine image of different models, I M =(1,2…M) T , M is the different code symbols of the engines of 7 types of 2.0L and 2.5L models, T is the number of times the camera scene group is photographed; Q0 is the unitary matrix, and C0 is the orthogonal space-time code.
4. The method for automatically inspecting engine finished parts according to claim 3, characterized in that: When the unitary matrix Q0 and the orthogonal space-time code C0 select a 7×7 model, we have: The ID reading matrix Q of the camera scene group and the encoding matrix C of the tag end are: In the formula, c1 is the engine code value of the first type of vehicle, c7 is the engine code value of the seventh type of vehicle, is the encoding value mapped in the orthogonal of the engine of the first model, It is the coding value mapped in the orthogonal of the engine of the 7th model.
5. The method for automatically inspecting engine finished parts according to claim 4, characterized in that: In the second step, the transmitted captured image signal of the camera scene group is encoded, and the expression is: R=(I M ×Q0)×(I M ×C0)H T +W In the formula, R is the captured image signal code, H T is the orthogonal space-time code, and W is the adjustment coefficient.
6. The method for automatically inspecting engine finished parts according to claim 2, characterized in that: The engine codes and image resolution of 2.0L and 2.5L models are obtained by analogy.
7. The method for automatically inspecting engine finished parts according to claim 1, characterized in that: In step S2, the fixed camera is changed into a movable camera using the trajectory motion control method built into the robot, and the robot is moved to reach the designated position, including: Step 1, obtaining a certain type or several types of trajectory motion vector mode camera trajectory change graphics, using a standard graphics library to convert the format, obtaining the corresponding image mapping format camera trajectory change graphics, and constructing a new data set using the trajectory motion vector mode camera trajectory change graphics and the image mapping format camera trajectory change graphics; Step 2: Use the convolutional neural network and attention module to obtain the global and local feature structures of the trajectory vectors in the new dataset; Step 3, the global and local feature structures of the trajectory route vector and the trajectory motion vector pattern camera trajectory change graph are input into a decoder composed of a long short-term memory network to obtain the trajectory route normal distribution parameter data and trajectory route state classification distribution probability data of the new data set; Step 4, randomly sampling from normal distribution to obtain the trajectory route position data of each scanning moving distance step, calculating the trajectory route state data of each scanning moving distance step from the trajectory route state classification step probability data, combining the output of each scanning moving distance step, and obtaining the combined predicted camera trajectory change graph; Step 5, inputting the generated camera trajectory change graph and the trajectory motion vector mode camera trajectory change graph into a discriminator composed of a long short-term memory network, obtaining deviation information of the discriminator predicting the input data, and obtaining a camera trajectory change graph generation model integrating a convolutional neural network and a long short-term memory network after training is completed; Step six, input the camera trajectory change graph of the trajectory motion vector pattern to be predicted and the corresponding image mapping format camera trajectory change graph into the camera trajectory change graph generation model integrating the convolutional neural network and the long short-term memory network for prediction, and obtain the final robot prediction generation camera trajectory change graph.
8. The method for automatically inspecting engine finished parts according to claim 1, characterized in that: In step S3, determining whether the finished engine component is consistent with the qualified product stored in the database includes: S301, obtaining response determination curve data of multiple groups of cameras for recognition of engine finished product parts; S302, extracting characteristic values of engine finished product components from the response determination curve; S303, preprocessing and standardizing the characteristic values to obtain a characteristic data set; S304, randomly dividing the feature data set into a training data set and a test data set; S305, using a deep learning network to train the training data set to obtain a comprehensive engine finished product component classification model, and using a test data set to test the comprehensive engine finished product component classification model, based on the test results and the qualified products stored in the database are compared to determine whether they are consistent, the classification types include 7 types of engines for 2.0L and 2.5L models.
9. An automatic inspection device for engine finished parts, characterized in that: The device is implemented by the automatic detection method for engine finished product parts according to any one of claims 1 to 8, and the device comprises: A camera scene group (1) is used to identify the engine model by using an ID reading method and to send a camera switching instruction to the camera scene group switching program; A robot (2) is used to convert a fixed camera into a movable camera by using a built-in trajectory motion control method, and utilize the movement of the robot to reach a specified position; A logic program judgment module (3) is used to judge the multiple groups of photographic results through a logic program to determine whether the engine finished product parts are consistent with the qualified products stored in the database; The communication display module (4) is used to output the determination result to the screen through communication means; and to save the detection data.
10. The automatic inspection device for engine finished parts according to claim 9, characterized in that: The engine finished product component automatic detection device is externally provided with a light-proof dark brown glass protective plate.
Citation Information
Cited By
Robot movement track generation method and system for engine detection
CN121468606A
