An intelligent fault detection and handling system for an injection molding machine

Abstract

The application belongs to the field of injection molding machines, and is an intelligent fault detection and processing system for an injection molding machine, wherein an AI interaction service module is arranged on the injection molding machine, a server includes a remote diagnosis module, the AI interaction service module and the remote diagnosis module are connected with a general control center through a network, and the system comprises the following steps: S1: a user inputs fault consultation content on the injection molding machine, and the AI interaction service module sends the fault consultation content to the server; S2: the general control center receives the fault consultation content and sends it to the remote diagnosis module; S3: the remote diagnosis module receives the fault consultation content, diagnoses the fault consultation content by calling a fault database and establishing a multi-modal fault diagnosis large model; S4: the remote diagnosis module sends the diagnosis result and a diagnosis scheme to the AI interaction service module through the general control center; and S5: the injection molding machine executes fault processing according to the diagnosis scheme, and the general control center saves the fault processing result to the fault database for updating and optimization.

Description

Technical Field

[0001] This invention belongs to the field of injection molding machines, and in particular, it is an intelligent fault detection and handling system for injection molding machines. Background Technology

[0002] Injection molding machine malfunctions are common in the injection molding industry, but sometimes the efficiency of handling them is low because the operators and maintenance personnel on site cannot handle them well. Therefore, there is an urgent need for an intelligent expert system to help users handle injection molding machine malfunctions efficiently.

[0003] Although existing expert systems are widely used, they only solve one type of fault and ignore the fact that in real-world scenarios, field workers may face injection molding machines or faults that exhibit multiple symptoms simultaneously.

[0004] Furthermore, the rule modules and inference engines in existing expert systems are often configured based on the computer's performance, and their operating efficiency may vary depending on the computer's configuration.

[0005] Therefore, there is an urgent need to develop an intelligent fault detection and handling system for injection molding machines that is suitable for various faults and can take into account the matching scheme between rule modules and inference engines under different computer conditions. Summary of the Invention

[0006] This invention proposes an intelligent fault detection and handling system for injection molding machines.

[0007] An intelligent fault detection and handling system for injection molding machines includes a user, an injection molding machine, a server, and a central control center. The injection molding machine is equipped with an AI interactive service module, and the server includes a remote diagnostic module. Both the AI ​​interactive service module and the remote diagnostic module are connected to the central control center via a network. The system includes the following steps:

[0008] S1: The user inputs fault inquiry content on the injection molding machine, and the AI ​​interactive service module sends the fault inquiry content to the server via the network;

[0009] S2: The central control center receives fault inquiries and sends them to the remote diagnostic module;

[0010] S3: The remote diagnostic module receives fault consultation content, matches the fault type, diagnoses the fault consultation content by retrieving the fault database and establishing a multimodal fault diagnosis model, and generates a diagnostic solution.

[0011] S4: The remote diagnostic module on the server side sends the diagnostic results and diagnostic solutions to the AI ​​interactive service module of the injection molding machine through the central control center;

[0012] S5: The injection molding machine performs fault handling according to the diagnostic plan and feeds back the fault handling results to the central control center. The central control center saves the fault handling results to the fault database for updating and optimization.

[0013] Preferably, the remote diagnostic module receives fault consultation content, matches the fault type, and diagnoses the fault consultation content by retrieving the fault database and establishing a multimodal fault diagnosis model, generating a diagnostic plan, which includes the following:

[0014] S31: Compare the fault consultation content received by the remote diagnostic module with historical fault resolution records and the fault database;

[0015] S32: If the comparison result is a known fault, then call the historical fault resolution records and fault database, and output the fault solution;

[0016] S33: If the comparison result is an unknown fault, the remote diagnostic module reads the status information of the injection molding machine, performs fault simulation and parameter analysis through the multimodal fault diagnosis model, generates corresponding parameter suggestions, and outputs fault solutions.

[0017] Preferably, the troubleshooting information includes two types of issues: inoperability and settings problems in the injection molding machine.

[0018] When an inoperable situation occurs, obtain the currently required operation object and send an AI consultation button malfunction consultation request to the central control center, which includes the malfunction consultation content.

[0019] When a setup problem occurs and the user does not know how to configure it, the name of the setup interface is obtained, and an AI consultation button malfunction consultation request is sent to the central control center.

[0020] The AI ​​interaction service module receives solutions and parameter suggestions from the central control center and performs troubleshooting and parameter adjustments.

[0021] As a preferred method, when an inoperable situation occurs, the system obtains the currently required operation object and sends an AI consultation button fault consultation request to the central control center, which includes the fault consultation content. The fault database contains operation object data corresponding to the interface screenshot data, and the fault database sends the operation object data to the injection molding machine's AI interaction service module.

[0022] As a preferred option, when a setup problem occurs and the user does not know how to configure it, the name of the setup interface is obtained, and an AI consultation button fault consultation request is sent to the central control center; the fault database contains interface data corresponding to the setup interface name, and the fault database sends the interface data to the AI ​​interaction service module of the injection molding machine.

[0023] Preferably, in step S1, the user inputs fault inquiry content on the injection molding machine, and the AI ​​interaction service module sends the fault inquiry content to the server via the network, including the following:

[0024] When a user sends a fault inquiry to the central control center, the fault inquiry content also includes a screenshot of the current injection molding machine interface;

[0025] After receiving the screenshot, the central control center will use the screenshot and the fault inquiry content to diagnose the problem through the fault database and the multimodal fault diagnosis model.

[0026] As a preferred option, when a user cannot find the parameter page or does not know how to adjust the parameters, the injection molding machine, after detecting that the user has triggered the AI ​​consultation button, sends a parameter adjustment consultation request to the server through the central control center; the server identifies the request content through a multimodal fault diagnosis big data model, provides guidance for the user to find the solution, and responds and adjusts according to the feedback.

[0027] As a preferred approach, when a user sends a request to the server from the injection molding machine, the server uses a multimodal fault diagnosis model to identify the current situation and searches through historical processing data to extract processing information under similar circumstances and provide feedback.

[0028] As a preferred option, when the server provides feedback to the injection molding machine, different feedback is given according to the user's permissions. Specifically, this includes: obtaining the user's permissions based on the authentication information of the user who sent the request; if the user is an operator, outputting the historical processing data of the current situation; and the operator processing the data based on the historical processing data and the on-site situation.

[0029] If the user is a tester, determine whether the user has the authority to directly perform parameter tuning and on-site processing:

[0030] If the user does not have the permission to directly adjust parameters, but only the permission to make suggestions, then parameter adjustment suggestions will be output and fed back to the injection molding machine and the user;

[0031] If the user has the authority to directly adjust parameters, then parameter adjustment can be performed on-site or remotely.

[0032] Preferably, the historical processing data includes the following: the specific scenario of the emergency, the handling process when facing the emergency, and the parameter adjustment information record.

[0033] The present invention has the following beneficial effects:

[0034] 1. This invention, based on AI, provides fault handling for injection molding machines. It can accurately and quickly report faults and provide timely solutions, offering better accuracy and immediacy in injection molding machine fault analysis. For fault detection, it helps identify existing faults in the injection molding machine and provides reference solutions based on these faults. Furthermore, it assists users in more direct parameter adjustment, utilizing AI's judgment capabilities to directly adjust parameters to specified positions, eliminating manual operation and improving the efficiency of parameter adjustment.

[0035] 2. This invention utilizes an AI consultation button to directly trigger AI diagnostic requests; simultaneously, it helps testers quickly adjust parameters by automatically performing parameter adjustments based on the tester's selections on the interface; it determines the tester's permissions based on their actual situation, directly adjusting parameters or providing adjustment suggestions; the automatic parameter adjustment allows for direct adjustment, saving time; and the precise parameter adjustment saves on adjustment costs.

[0036] 3. This invention utilizes an AI-powered consultation button to modify injection molding machine parameters; it assists testing personnel in correcting parameter modifications; it provides modification or parameter adjustment suggestions for unexpected problems or malfunctions on-site; simultaneously, by using AI to detect problems, parameter adjustments are more accurate, saving manual adjustment time, avoiding errors in manual parameter adjustment, improving adjustment efficiency, and preventing parameter errors or setting mistakes.

[0037] 4. The AI-powered consultation content in this invention can respond to questions raised by operators and return corresponding operational suggestions or suggested results based on user permissions, thus achieving AI interaction. The AI-powered consultation content can also provide modification suggestions to operators or make direct modifications based on user permissions. Attached Figure Description

[0038] Figure 1 This is a schematic diagram of the steps of an intelligent fault detection and handling system for injection molding machines according to the present invention;

[0039] Figure 2 This is a schematic diagram of the structure of an intelligent fault detection and handling system for injection molding machines according to the present invention. Detailed Implementation

[0040] To enable those skilled in the art to better understand the technical solutions of this invention, the technical solutions in the embodiments of this invention will be clearly described below in conjunction with the examples.

[0041] like Figure 1 As shown, this invention proposes an intelligent fault detection and handling system for injection molding machines, comprising the following steps:

[0042] (1) Set at least one AI consultation button on the injection molding machine; each button corresponds to a specific fault consultation scenario and appears in the form of a pop-up dialog box;

[0043] (2) The server has a fault scenario matching and registration model based on deep learning that is connected to the AI ​​consultation button. After the AI ​​consultation button is triggered, the fault consultation content corresponding to the button is sent to the server via the network.

[0044] The server also includes a remote diagnostic module connected to the central control center, which in turn connects to the injection molding machine via a network, using methods such as TCP / IP and UDP. When the AI ​​consultation button is triggered, the corresponding fault consultation content, including the fault consultation request, is sent to the server via the network for fault diagnosis and parameter adjustment.

[0045] The method includes the following steps: After the AI ​​consultation button is triggered via the network, the central control center sends the corresponding fault consultation content to the remote diagnostic module on the server. The remote diagnostic module receives the fault consultation content and performs corresponding diagnosis, using a fault matching degree calculation formula during the diagnosis process. ,in The overall fault matching degree is measured in ranges from 0 to 1, with values ​​closer to 1 indicating a higher matching degree. This sets the weight for text matching; the default value is 0.6, and it can be dynamically adjusted based on the fault type. The cosine similarity between the fault consultation text and historical fault text is used. Set weights for the image (default 0.4). The feature similarity between the interface screenshot and the historical fault screenshot is calculated after extracting features using a convolutional neural network; if the fault is known, If the fault is unknown, the historical fault resolution record will be retrieved, the fault solution will be output, and sent to the injection molding machine for fault solution selection or fault result confirmation, and finally the solution will be selected or confirmed; The remote diagnostic module remotely controls the injection molding machine's control system, reads the machine's status information, performs fault simulation and parameter analysis, and generates parameter suggestions corresponding to unknown faults. The suggested parameter values ​​are based on... Calculation, where To suggest adjusting the parameter values, This is the real-time collected value of the current parameter. For correction factor, Depending on the matching degree Dynamic changes For historical similar faults The average parameter adjustment amount is sent to the injection molding machine for debugging until the debugging is completed.

[0046] An AI interactive service module is installed on the injection molding machine. This module is connected to the central control center via the network. When the user triggers the AI ​​consultation button on the injection molding machine, the module sends an AI consultation button fault consultation request to the central control center via the network.

[0047] The fault consultation request contains fault consultation content; the fault consultation content includes situations where it is impossible to operate or how to set it up; if it is impossible to operate, the current required operation object is obtained, and then an AI consultation button fault consultation request is sent to the central control center; if it is unknown how to set it up, the name of the settings interface is obtained and an AI consultation button fault consultation request is sent; the AI ​​interaction service module receives solutions or parameter suggestions from the central control center, and then performs troubleshooting or parameter adjustment.

[0048] When a user sends a fault inquiry to the central control center, they take a screenshot of the current injection molding machine interface and send it as an attachment to the central control center. Upon receiving the attachment, the central control center inputs the current AI module and the AI ​​inquiry request into the fault database or a Transformer-based multimodal fault diagnosis model for diagnosis. If the fault is known, the fault database outputs a fault solution and sends it as an attachment to the injection molding machine's AI interaction service module. If the fault is unknown, the Transformer-based multimodal fault diagnosis model performs fault simulation and parameter analysis, and the central control center sends the obtained parameter suggestions as an attachment to the injection molding machine's AI interaction service module.

[0049] The central control center houses a fault database and a large-scale multimodal fault diagnosis model based on Transformer. The fault database stores fault types and corresponding solutions, specifically including:

[0050] (1) Insufficient injection: The fault feature text is "Insufficient injection volume, product lack of material", and the feature screenshot shows that the injection position curve is below the threshold. The solution is to increase the injection pressure to 120 bar and extend the injection time by 0.5 s.

[0051] (2) Abnormal temperature: The fault feature text is "the nozzle temperature is too high and scorch marks appear", and the feature screenshot shows that the temperature exceeds 250°C. The solution is to reduce the nozzle temperature to 230°C and clean the nozzle.

[0052] (3) Mold not closing tightly: The fault feature text is "large mold closing gap, with overflow", and the feature screenshot shows that the mold closing position deviation is >0.5mm. The solution is to adjust the mold closing force to 300kN and calibrate the mold closing sensor.

[0053] (4) Insufficient oil pressure: The fault characteristic text is "low system pressure, slow operation", and the characteristic screenshot shows that the pressure gauge displays <80 bar. The solution is to check the hydraulic pump, add hydraulic oil to the level line, and replace the filter.

[0054] (5) Poor cooling: The fault characteristic text is "Insufficient product cooling, severe deformation", and the characteristic screenshot shows that the cooling water temperature is >35℃. The solution is to increase the cooling water flow rate to 5L / min and clean the cooling pipes.

[0055] When the fault solution is sent as an attachment to the AI ​​interactive service module of the injection molding machine, the AI ​​interactive service module first obtains the current user's operation permissions. If the current user is a tester, the module will directly operate according to the obtained fault solution. If the current user is an operator, the module will select a fault solution or confirm the fault result according to the fault solution, and finally select or confirm the solution. If the current user has operation permissions, the module will directly implement the fault solution or adjust the parameters.

[0056] The fault consultation content is "unable to operate". The corresponding fault consultation request includes: the current interface screenshot data; the fault database contains the operation object data corresponding to the interface with the current interface screenshot data; the fault database sends the operation object data as an attachment to the injection molding machine's AI interaction service module.

[0057] The fault inquiry content is "I don't know how to set it up". The corresponding fault inquiry request includes: setting the interface name; having interface data corresponding to the setting interface name in the fault database; and sending the interface data as an attachment to the injection molding machine's AI interaction service module.

[0058] A method for intelligent fault detection and handling of injection molding machines. The method involves an injection molding machine that communicates with a server and has an AI consultation button on the machine. Specific steps include: detecting when a user triggers the AI ​​consultation button; sending a parameter adjustment consultation request to the server, the request including specific scenarios where the user cannot find the parameter page or is unsure how to adjust parameters in unexpected situations; the server using a Transformer-based multimodal fault diagnosis model to identify the request content; if the request indicates a missing parameter page, remotely controlling the injection molding machine to provide guidance to the user; and responding and adjusting based on further user feedback.

[0059] The parameter tuning consultation request includes user permission information and the specific scenario for handling emergencies. The specific scenario for handling emergencies is obtained from historical parameter tuning records. If the user permission is remote control, parameter tuning is performed directly; if the user permission is only suggestion, parameter tuning suggestions are output, which are based on formulas. Calculate; take appropriate action based on the user's response.

[0060] The server uses a Transformer-based multimodal fault diagnosis model to identify the current interface and search historical parameter tuning records for records with the same interface, extracting records for handling emergencies. If no matching interface is found, suggestions are provided to the user, who confirms and the suggestions are sent. If no suggestion is selected, the process ends and a request to end is output. If a suggestion is selected, the extracted results are fed back to the user. If the user has remote control permissions, parameter tuning is performed directly. If the user only has suggestion permissions, parameter tuning suggestions are output. The server then takes appropriate action based on the user's response.

[0061] The user request contains interface identification information, which is obtained by identifying the current interface of the injection molding machine.

[0062] The user request contains interface recognition information, which is obtained by recognizing the current interface of the injection molding machine.

[0063] The server uses a Transformer-based multimodal fault diagnosis model to identify the current interface and search historical parameter tuning records for records with the same interface, extracting records for handling emergencies. If no matching interface is found, suggestions are provided to the user, who confirms and the suggestions are sent. If no suggestion is selected, the process ends and a request to end is output. If a suggestion is selected, the extracted results are fed back to the user. If the user has remote control permissions, parameter tuning is performed directly. If the user only has suggestion permissions, parameter tuning suggestions are output. The server takes appropriate actions based on the user's response. User permissions are obtained through the user's authentication information.

[0064] User permissions are obtained through user authentication information.

[0065] The server uses a Transformer-based multimodal fault diagnosis model to identify the current interface and search historical parameter tuning records for records with the same interface, extracting records for handling unexpected situations. If no matching interface is found, suggestions are output to the user, who confirms and the suggestions are sent. If no suggestion is selected, the process ends and a request termination signal is output. If a suggestion is selected, the extracted results are fed back to the user. If the user in the request is a tester, and their permissions allow remote control, parameter tuning is performed directly. If their permissions only allow suggestion, parameter tuning suggestions are output. The server takes appropriate actions based on the user's response. User permissions are obtained through the user's authentication information.

[0066] The historical parameter tuning records include records of situations encountered during emergencies. These records include: the specific scenario of the emergency, the handling process, and parameter tuning information. A specific example is as follows:

[0067] (1) Time 08:30, temperature 220℃, injection position 50mm, machine number M012, parameter adjustment information is that the injection pressure is adjusted from 100bar to 110bar, the process is to solve the problem of insufficient injection and product material shortage.

[0068] (2) Time 10:15, temperature 240℃, injection position 48mm, machine number M005, parameter adjustment information is that the nozzle temperature drops from 240℃ to 220℃, and the processing is to eliminate the scorch marks on the product surface.

[0069] (3) Time 14:15, temperature 235℃, injection position 45mm, machine number M008, parameter adjustment information is to extend the holding time from 2s to 2.5s, and the processing is to solve the problem of internal air bubbles in the product;

[0070] (4) Time 16:40, temperature 225℃, injection position 52mm, machine number M012, parameter adjustment information is that the clamping force is adjusted from 280kN to 310kN, and the processing is to solve the overflow problem caused by poor clamping.

[0071] (5) Time 18:20, temperature 230℃, injection position 49mm, machine number M003, parameter adjustment information is to increase the cooling water flow rate from 3L / min to 6L / min, and the processing is to solve the deformation problem caused by insufficient product cooling.

[0072] The specific scenarios for handling emergencies include time, temperature, injection location, machine number, and corresponding parameter adjustment information. These scenarios are pre-set by the server or modified and saved by the user in the historical parameter adjustment record.

[0073] An intelligent parameter adjustment system for injection molding machines includes: a module for collecting user AI consultation button requests; a module for collecting user authentication information; a module for responding to AI consultation button requests and judging the content of the response; a module for judging user permissions; if the user's permissions are limited to suggestion only, then output parameter adjustment suggestion information; and a module for taking corresponding actions based on the user's response.

[0074] The parameter tuning suggestions can be provided to the user through the interface for confirmation and execution; or the parameter tuning suggestions can be executed remotely.

[0075] Example 1:

[0076] This solution relates to an intelligent fault detection and handling system for injection molding machines. It features an AI-powered quick-access button on the injection molding machine. When encountering a problem, users can directly click the button to interact with the backend database, a Transformer-based multimodal fault diagnosis model, and human assistance to obtain relevant solutions. Simultaneously, when a fault occurs, users can send fault notifications to the operator / test user through the server and central control terminal to assist in handling the issue.

[0077] This solution is applicable to situations involving the detection and handling of injection molding machine malfunctions.

[0078] Combination Figure 2 The diagram shows the structure of the injection molding machine connecting to the server, where the injection molding machine can communicate with the server via wired / wireless network.

[0079] At least one AI consultation button is set on the injection molding machine as the entry point for users to trigger AI; each button corresponds to a specific fault consultation scenario, which appears in the form of a pop-up dialog box.

[0080] Troubleshooting buttons are located in appropriate positions on the injection molding machine, including the following:

[0081] (1) If the user cannot find the parameter page or does not know how to adjust the parameters in the face of an emergency, they can directly set a "Parameter Adjustment Consultation" button on the injection molding machine. Clicking it will bring up a dialog box. The user should fill in the interface name, such as time, temperature, injection position, or machine number including the injection cylinder, and then submit it. The parameter adjustment consultation request, including the interface identification information, will be sent to the remote diagnostic module via the network. The remote diagnostic module will retrieve the historical parameter adjustment records, match and identify them according to the interface identification information / machine number, and extract the matching results. The matching process uses the fault matching degree calculation formula. The results are then fed back to the user. The matching methods in this solution include character matching and similarity matching; for example, character matching can be used for time, temperature, and injection location, while machine numbers can be directly compared.

[0082] (2) If users encounter problems with operation or do not know how to set it up, they can also click the shortcut button for help and enter instructions in the pop-up box. For example, if they cannot operate or do not know how to set it up, they can also make an explanation, take a screenshot on the injection machine UI, set an AI consultation button on the injection molding machine, and after the AI ​​consultation button is triggered, the corresponding fault consultation content will be sent to the server via the network.

[0083] In this solution, the server pre-sets some fault scenarios; for example, the following are set: (1) For scenarios where operation is impossible, after the AI ​​consultation button is triggered, the server obtains the corresponding interface image, performs content judgment, and determines the operation object that the user wants to perform; for example, "click cannot be operated", after the AI ​​consultation button is triggered, the server obtains the corresponding interface image, identifies the operation object that the user needs to click, obtains the icon image of the corresponding function key, calls the remote diagnostic module, obtains the interface data of the same operation object in the historical parameter adjustment record, and sends it to the injection molding machine; after the injection molding machine receives the interface data of the same operation object in the historical parameter adjustment record, a pop-up window appears: A. prompt operation, after clicking OK, it will be automatically called and the result will be sent to the server; B. perform directly, no prompt is needed, and the result will be sent to the server.

[0084] (2) If users encounter problems with operation or do not know how to set it up, they can also click the shortcut button for help, enter instructions in the pop-up box, set the AI ​​consultation button on the injection molding machine, and send the corresponding fault consultation content to the server via the network after the AI ​​consultation button is triggered.

[0085] The server obtains a screenshot of the corresponding interface, performs content analysis, and determines whether the user is unable to perform the operation or does not know how to set it up.

[0086] An AI consultation button is set on the injection molding machine. When the AI ​​consultation button is triggered, the corresponding fault consultation content is sent to the server via the network.

[0087] (3) For cases where the user does not know how to set it up, after the AI ​​consultation button is triggered, the server obtains the interface content of the corresponding settings interface description, performs content judgment and filtering, and determines what the user wants to set up.

[0088] For example, if a user inputs "adjust parameters" or "temperature increases," and the AI ​​consultation button is triggered, the server retrieves the corresponding settings interface description "temperature high / low." If the interface corresponding to the settings interface description "temperature high / low" is indeed a settings interface, the remote diagnostic module is invoked to retrieve similar records from historical parameter adjustment records and check the fault matching rate. judge, For known faults, send parameter suggestions, with suggested parameter values ​​according to... The system calculates parameters, such as "lower to 180℃", and sends the results to the injection molding machine. After receiving the parameter suggestions, the injection molding machine displays a pop-up window.

[0089] A. The user is prompted to click OK to register the corresponding parameters. For example, the interface for "temperature level" is the settings interface. When the user clicks to send the message, "lower to 180℃" will be automatically registered, and the result will be sent to the server.

[0090] B. Ask if you accept the parameter suggestions. Click OK to send the data modification information to the server.

[0091] (4) For other or complex situations, after the AI ​​consultation button is triggered, the server obtains the corresponding interface content, remotely controls the machine to perform remote diagnosis, obtains parameter suggestions, and sends the results to the injection molding machine. For example, "lower", after the injection molding machine receives the information, it prompts: A. prompts operation, clicks OK, automatically calls and sends the results to the server; B. performs directly without prompting, and sends the results to the server.

[0092] (5) When encountering other or complex situations, users can also click the shortcut button for prompt help and enter instructions in the pop-up box. The instructions directly include the interface or parameters that need to be adjusted. Set up an AI consultation button on the injection molding machine. After the AI ​​consultation button is triggered, the corresponding fault consultation content will be sent to the server via the network.

[0093] (6) When a fault occurs, users can send fault notifications to the operator / test user through the server and the central control terminal to assist them in handling the fault. For example, for a fault in the electronic control module, it can be sent to the test team. After the testers see the fault information, they can remotely debug and control it through the operation control. For example, when a cylinder leak occurs, the central control center sends a fault notification to the operator and sends the specific fault parameter information to the user when necessary.

[0094] In order to obtain user permissions, registration and authentication information is set in the AI ​​interaction service module.

[0095] The authentication information includes the following:

[0096] (1) Operator information;

[0097] (2) Current interface information of the operator;

[0098] (3) After the operator enters the identity information at the authentication terminal, feedback is provided on the current interface recognition. If the current interface recognition information is the same as the operator's current interface information, the authentication is successful and parameter adjustment and remote debugging can be performed. If the current interface recognition information is not the same as the operator's current interface information, the authentication is unsuccessful and parameter adjustment suggestions can only be provided on the interface and sent to the operator.

[0099] The parameter tuning recommendations are determined based on historical parameter tuning data from historical tuning records, using a formula. The calculations show that the historical parameter tuning records specifically include:

[0100] (1) Time 08:30, temperature 220℃, injection position 50mm, machine number M012, parameter adjustment information is that the injection pressure is adjusted from 100bar to 110bar, the process is to solve the problem of insufficient injection and product material shortage.

[0101] (2) Time 10:15, temperature 240℃, injection position 48mm, machine number M005, parameter adjustment information is that the nozzle temperature drops from 240℃ to 220℃, and the processing is to eliminate the scorch marks on the product surface.

[0102] (3) Time 14:15, temperature 235℃, injection position 45mm, machine number M008, parameter adjustment information is to extend the holding time from 2s to 2.5s, and the processing is to solve the problem of internal air bubbles in the product;

[0103] (4) Time 16:40, temperature 225℃, injection position 52mm, machine number M012, parameter adjustment information is that the clamping force is adjusted from 280kN to 310kN, and the processing is to solve the overflow problem caused by poor clamping.

[0104] (5) Time 18:20, temperature 230℃, injection position 49mm, machine number M003, parameter adjustment information is to increase the cooling water flow rate from 3L / min to 6L / min, and the processing is to solve the deformation problem caused by insufficient product cooling.

[0105] When a user clicks "suggestion send," the specific scenario and parameter adjustment information for the emergency will be recorded and sent to the operator. The specific scenario, handling process, and parameter adjustment information for the emergency will also be stored in the remote diagnostic module.

[0106] After receiving parameter tuning suggestions, users can receive them through the following interface:

[0107] (1) The operator conducts the test;

[0108] (2) The operator performs remote debugging;

[0109] (3) After adjusting the parameters, ask whether to confirm. After confirmation, return the confirmation result to the central control center.

[0110] Meanwhile, the server can determine whether to perform parameter tuning based on historical parameter tuning records and the query for parameter tuning suggestions. Specifically, this includes: A: If the specific scenario and parameter tuning information records can fully cover the unexpected situation, If so, parameter adjustment suggestions will be sent directly. After the user confirms, the confirmation result will be sent to the central control center.

[0111] B: If the specific scenario and parameter adjustment information records cover the unexpected situation. The system obtains the specific scenario of the emergency and sends it to the operator. The operator then confirms the scenario and sends the result to the central control center.

[0112] C: If the specific scenario and parameter adjustment information are not recorded in the case of an emergency, then... If the user requests a parameter adjustment suggestion, the system will ask whether to send a message. If the user requests a message, the system will send the on-site parameter adjustment results to the central control center. If the user requests a message, the system will send the on-site parameter adjustment results, the specific scenario of the emergency, and the parameter adjustment information record to the central control center so that the system can adjust the parameters directly in the future if the same emergency occurs.

[0113] After receiving the parameter tuning suggestions, users can click "Confirm." Once the suggestions are confirmed, the confirmation result will be sent to the central control center. If the user does not confirm, the process will be handled by the operator or a remote device.

[0114] At the same time, the central control center sends fault notifications to the operators. The central control center is part of the server and is connected to the injection molding machine through the network. The connection methods include TCP / IP, UDP, etc. Among them, TCP / IP is connected to the central control center, and the central control center is connected to the injection molding machine through the network.

[0115] This invention features an AI consultation button on the injection molding machine. When a user clicks the button, a dialog box pops up. Users can directly input fault-related questions via the input box and send them to the central control center; alternatively, they can input text, voice, or screenshots. The injection molding machine will then retrieve the voice input or screenshot content and send it to the central control center.

[0116] The central control center has a large multimodal fault diagnosis model based on Transformer, a remote diagnosis module, and an AI interactive service module that receives solutions or parameter suggestions from the central control center and then performs fault troubleshooting or parameter adjustment.

[0117] Example 2

[0118] This embodiment uses the intelligent detection and handling of "unstable injection" fault in injection molding machines as an example to explain in detail the implementation process of this system:

[0119] Step 1: Fault Triggering and Information Collection. During production using an injection molding machine (model M015), the operator discovers fluctuations in injection volume leading to material shortages or flash on the product. At this point, pressing the "Injection Fault" AI consultation button on the injection molding machine's control panel will bring up a dialog box. The operator enters "Unstable injection, product material shortage and occasional flash" in the dialog box and clicks the "Screenshot Upload" button. The current injection parameters interface displays a current temperature of 225℃, injection pressure of 105 bar, and injection position of 48mm. The screenshot is submitted as an attachment.

[0120] Step 2: Information Transmission and Diagnosis Initiation. The injection molding machine's AI interactive service module sends the text and screenshots of the fault consultation content to the central control center via TCP / IP protocol. The central control center forwards them to the remote diagnosis module on the server. The remote diagnosis module calls the Transformer-based multimodal fault diagnosis model to diagnose the received information.

[0121] Step 3: Fault Matching and Judgment. The large model first calculates the fault matching degree. ,in , ; The cosine similarity between the input text and historical "instability in injection" error text was calculated to be 0.85. The feature similarity between the current screenshot and historical fault screenshots was calculated to be 0.78, and the final result was... ,because This is determined to be a known fault.

[0122] Step 4: Solution Generation and Feedback. The remote diagnostic module retrieves the solution with the highest matching degree from the fault database. The historical solution was: "1. Reduce the injection temperature to 215℃; 2. Increase the injection pressure to 115 bar; 3. Adjust the injection position to 52mm", and this solution was sent to the injection molding machine's AI interactive service module through the central control center.

[0123] Step 5: Permission Verification and Solution Execution. The AI ​​interaction service module detects that the current user is an operator with "suggestion only" permission. A solution pop-up window appears on the interface. After the operator confirms the solution, they click "Execute Suggestion," and the injection molding machine automatically adjusts its parameters according to the solution. After adjustment, 10 molded products are continuously monitored. If the injection is stable, the operator clicks "Confirm Solution," and the result is sent to the central control center for archiving.

[0124] Step 6: Handling Unknown Faults (Hypothetical Scenario). If the calculation in Step 3 yields... ( The fault was determined to be unknown. The remote diagnostic module remotely reads the real-time status data of the injection molding machine, including hydraulic pressure fluctuation curves and motor current, performs fault simulation, and determines the fault based on the formula. Calculation parameter tuning recommendations, among which The current injection pressure is 105 bar. , The average adjustment for similar faults is 12 bar, therefore... The value is rounded down to 116 bar, and temperature and position adjustment suggestions are generated simultaneously. These suggestions are sent to the injection molding machine. The testing personnel, with "remote control" permissions, confirm and remotely execute the debugging. Repeat the fine-tuning three times until the fault is resolved. The debugging process data is automatically saved to the historical parameter adjustment record.

[0125] It should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An intelligent fault detection and handling system for injection molding machines, characterized in that, The system includes a user, an injection molding machine, a server, and a central control center. The injection molding machine is equipped with an AI interactive service module, and the server includes a remote diagnostic module. Both the AI ​​interactive service module and the remote diagnostic module are connected to the central control center via a network. The system includes the following steps: S1: The user inputs fault inquiry content on the injection molding machine, and the AI ​​interactive service module sends the fault inquiry content to the server via the network; S2: The central control center receives fault inquiries and sends them to the remote diagnostic module; S3: The remote diagnostic module receives fault consultation content, matches the fault type, diagnoses the fault consultation content by retrieving the fault database and establishing a multimodal fault diagnosis model, and generates a diagnostic solution. S4: The remote diagnostic module on the server side sends the diagnostic results and diagnostic solutions to the AI ​​interactive service module of the injection molding machine through the central control center; S5: The injection molding machine performs fault handling according to the diagnostic plan and feeds back the fault handling results to the central control center. The central control center saves the fault handling results to the fault database for updating and optimization. The troubleshooting information includes two types of issues: inoperability and settings problems in the injection molding machine. When an inoperable situation occurs, obtain the currently required operation object and send an AI consultation button malfunction consultation request to the central control center, which includes the malfunction consultation content. When a setup problem occurs and the user does not know how to configure it, the name of the setup interface is obtained, and an AI consultation button malfunction consultation request is sent to the central control center. The AI ​​interaction service module receives solutions and parameter suggestions from the central control center and performs troubleshooting and parameter adjustments. When an inoperable situation occurs, the system retrieves the currently required operation object and sends an AI consultation button fault consultation request to the central control center, which includes the fault consultation content. The fault database contains operation object data corresponding to the interface screenshot data. The fault database then sends the operation object data to the injection molding machine's AI interaction service module.

2. The intelligent fault detection and handling system for injection molding machines according to claim 1, characterized in that, The remote diagnostic module receives fault inquiries, matches the fault type, and diagnoses the inquiries by retrieving fault databases and establishing a multimodal fault diagnosis model. It then generates a diagnostic plan, which includes the following: S31: Compare the fault consultation content received by the remote diagnostic module with historical fault resolution records and the fault database; S32: If the comparison result is a known fault, then call the historical fault resolution records and fault database, and output the fault solution; S33: If the comparison result is an unknown fault, the remote diagnostic module reads the status information of the injection molding machine, performs fault simulation and parameter analysis through the multimodal fault diagnosis model, generates corresponding parameter suggestions, and outputs fault solutions.

3. The intelligent fault detection and handling system for injection molding machines according to claim 1, characterized in that, When a setup problem occurs and the user does not know how to configure it, the name of the setup interface is obtained, and an AI consultation button malfunction consultation request is sent to the central control center. The fault database contains interface data corresponding to the settings interface name, and the fault database sends the interface data to the injection molding machine's AI interaction service module.

4. The intelligent fault detection and handling system for injection molding machines according to claim 1, characterized in that, In step S1, the user inputs fault inquiry content on the injection molding machine, and the AI ​​interaction service module sends the fault inquiry content to the server via the network, including the following: When a user sends a fault inquiry to the central control center, the fault inquiry content also includes a screenshot of the current injection molding machine interface; After receiving the screenshot, the central control center will use the screenshot and the fault inquiry content to diagnose the problem through the fault database and the multimodal fault diagnosis model.

5. The intelligent fault detection and handling system for injection molding machines according to claim 1, characterized in that, When a user cannot find the parameter page or does not know how to adjust the parameters, the injection molding machine, after detecting that the user has triggered the AI ​​consultation button, sends a parameter adjustment consultation request to the server through the central control center. The server identifies the request content through a multimodal fault diagnosis model, provides guidance to the user, and responds and adjusts according to the feedback.

6. The intelligent fault detection and handling system for injection molding machines according to claim 1, characterized in that, When a user sends a request to the server from the injection molding machine, the server uses a multimodal fault diagnosis model to identify the current situation and searches through historical processing data to extract processing information under the same conditions and provide feedback.

7. The intelligent fault detection and handling system for injection molding machines according to claim 6, characterized in that, When the server provides feedback to the injection molding machine, different feedback is given according to the user's permissions. Specifically, the user's permissions are obtained based on the authentication information of the user who sent the request. If the user is an operator, the historical processing data of the current situation is output, and the operator processes the data based on the historical processing data and the on-site situation. If the user is a tester, determine whether the user has the authority to directly perform parameter tuning and on-site processing: If the user does not have the permission to directly adjust parameters, but only the permission to make suggestions, then parameter adjustment suggestions will be output and fed back to the injection molding machine and the user; If the user has the authority to directly adjust parameters, then parameter adjustment can be performed on-site or remotely.

8. The intelligent fault detection and handling system for injection molding machines according to claim 6, characterized in that, The historical processing data includes the following: the specific scenarios of emergencies, the handling process when facing emergencies, and parameter adjustment information records.

Images