An eutectic patch machine full-scene operation abnormality intelligent alarm device and method

Abstract

The application discloses a kind of eutectic pasting machine full scene operation exception intelligent alarm device and method, device includes with eutectic pasting machine full scene operation external equipment sequentially connected full scene exception signal acquisition module, exception alarm main control module and system main control module, wherein full scene exception signal acquisition module, for collecting the equipment state signal of external equipment, according to equipment state signal to obtain exception signal and the ID number of exception equipment corresponding with exception signal and exception number;Exception alarm main control module, for obtaining the exception detailed information corresponding with exception number;System main control module, for according to interrupt signal to eutectic pasting machine full scene operation external equipment is carried out emergency exception reset operation and starts buzzer alarm, solve the prior art cannot carry out accurate monitoring and rapid alarm to the technical problem that eutectic pasting machine operates abnormally, realizes the accurate comprehensive judgment of exception signal and the rapid acquisition of exception detailed information.

Description

TECHNICAL FIELD

[0001] The present application mainly relates to the technical field of eutectic patch, and particularly relates to a full-scene operation abnormality intelligent alarm device and method of eutectic patch machine. BACKGROUND

[0002] The eutectic patch machine is a key equipment for realizing full-automatic chip welding by combining a mechanical arm, a mechanical claw vacuum suction nozzle, a eutectic heating table and a camera and the like. The diversity of components of the eutectic patch machine increases the possibility of abnormality, which will affect the high-quality production of the chip. Therefore, in order to ensure the error-free joint operation of the mechanical arm, the mechanical claw suction nozzle and the camera and the like, it is necessary to design a reliable abnormality alarm system. The present application provides a full-scene operation abnormality intelligent alarm of eutectic patch machine, which can monitor the operation of each component of the eutectic patch machine in real time, accurately judge abnormality, write system error logs, alarm and remind the staff, and take timely corresponding control measures on the component with abnormality to prevent the abnormality from causing system operation collapse and to minimize the economic loss caused by the abnormality of the component.

[0003] At present, the design of the abnormality intelligent alarm system is mainly divided into two types of based on a single type of sensor and based on multiple types of sensors. The abnormality intelligent alarm system based on a single type of sensor often selects a sensor most suitable for the monitoring object and the monitoring demand as the abnormality detection device according to the monitoring object and the monitoring demand. This kind of system has the advantages of simple construction and low cost, but is not suitable for system objects with multiple components and complex processes. The abnormality intelligent alarm system based on multiple types of sensors uses multiple sensors to monitor the object system in real time from multiple angles. The abnormality intelligent alarm system constructed by multiple types of sensors has faster abnormality alarm speed and more accurate system abnormality type judgment. Meanwhile, based on the monitoring demand of multiple components of the eutectic patch machine, the present application adopts multiple types of sensors to construct the abnormality intelligent alarm system.

[0004] The utility model discloses CN201999614U patent is elevator hall door abnormal opening intelligence alarm system. The intelligence alarm system includes hall door infrared sensor, car infrared sensor, external input processing circuit, wireless radio frequency transmitting circuit, alarm drive circuit and central processing unit etc. Among them, the infrared sensor of hall and car is connected with external input processing circuit. Central processing unit combines the floor number information that external input processing circuit and floor address code circuit input, and the signal is output to wireless radio frequency transmitting circuit, serial communication interface circuit and alarm drive circuit, when elevator door abnormal opening, the person of the person who wants to enter the elevator is prewarned, and the life safety of the person who takes the elevator is guaranteed. The system utilizes two infrared sensors, and the opening situation of elevator door is obtained in time through the central processing unit processing sensor result, and the occurrence of elevator accident is avoided to the maximum extent. But this system only utilizes a kind of sensor, for eutectic chip mounter full scene operation abnormality detection, sensor is too single, and part, such as mechanical arm's operation abnormality detection, infrared sensor is not applicable. Therefore, this alarm system is not applicable to eutectic chip mounter full scene operation abnormality detection.

[0005] The patent CN111862547A application is an intelligent alarm system based on power supply cabinet internal abnormality. The system is composed of a collection unit, a first connection module, a comparison module, a database, a processor, an output module, an execution module, a volume up and down module, and a PC and mobile terminal alarm module. The collection unit includes a temperature collection module, a PC terminal position noise collection module and a mobile terminal position noise collection module. The intelligent alarm system can automatically adjust the alarm volume according to environmental factors. The intelligent alarm system provides alarm information for workers, ensuring timely maintenance of the power supply cabinet. At the same time, through the PC terminal position noise collection module and the mobile terminal position noise collection module, the alarm volume is adjusted according to the size of the environmental noise on site, so that the alarm volume is moderate and the comfort of the on-site workers is improved. This system uses multiple sensors to achieve the function of intelligently adjusting the alarm volume according to the environmental noise of the work site. However, this system only realizes the sound alarm function when the abnormality occurs, and there is no detailed introduction of the timely handling of the abnormality afterwards. After the alarm, the abnormal part needs to be operated immediately, such as power off and return, to further reduce the damage to the device caused by the abnormality. The eutectic chip mounter is a key device for chip welding, and the emergency control processing after the configuration abnormality occurs can minimize the damage to the chip, so this system cannot be used to build a full-scene operation abnormality intelligent alarm system for eutectic chip mounter.

[0006] The utility model discloses a patent announcement number CN215265050U practical new type patent is a kind of for wastewater treatment's abnormal alarm system, and the system is composed of data acquisition device, main control system and drive device.Data acquisition device includes sewage quality detector, flow sensor, ultrasonic pipeline leak detector, positioning device, water pressure sensor, and above-mentioned sensor is connected with second, third, fourth, fifth, sixth wireless communication device respectively.Second to sixth wireless communication device and first wireless communication device also establish communication connection.Main control system includes first wireless communication device, man-machine interface, data storage device, alarm device, clock device and central controller.Drive device includes wastewater drainage pipe electric valve driver, electric flow regulating valve driver, drainage pump power regulator, wastewater drainage pipe electric valve, electric flow regulating valve, drainage pump.This system realizes the judgment whether pipeline is broken and the detection of the specific position of breakage, simultaneously realizes the automatic regulation of flow and water pressure in pipe using drive device.The system utilizes a variety of sensors, detects wastewater treatment pipeline from multiple aspects, and also designs corresponding drive device to cope with pipeline breakage and other abnormalities, control pipeline water pressure and flow, and reduce the impact of pipe breakage on the overall process of wastewater treatment.But when facing multiple signals transmitted by multiple sensors, the system does not specify the specific processing method of multiple signals.Eutectic patch machine full-scene operation abnormality is various, and the number of required sensors is numerous.According to the information transmitted by numerous sensors, it is judged whether there is an abnormality, and which abnormality is the premise of realizing eutectic patch machine full-scene operation abnormality intelligent alarm, therefore the system lacking intelligent fusion judgment of multiple sensor information cannot be applied to eutectic patch machine full-scene operation abnormality intelligent alarm system. SUMMARY

[0007] The eutectic patch machine full-scene operation abnormality intelligent alarm device and method provided by the application solve the technical problem that the prior art cannot accurately monitor and quickly alarm the operation abnormality of the eutectic patch machine.

[0008] To solve the above technical problems, the eutectic patch machine full-scene operation abnormality intelligent alarm device provided by the application comprises, in sequence, a full-scene abnormal signal acquisition module, an abnormal alarm master control module and a system master control module connected with an eutectic patch machine full-scene operation external device, wherein:

[0009] The full-scene abnormal signal acquisition module is used to acquire the device state signal of the eutectic patch machine full-scene operation external device, obtain the abnormal signal according to the device state signal, and obtain the ID number and abnormal number of the abnormal device corresponding to the abnormal signal.

[0010] The abnormal alarm master control module is used to obtain the abnormal detailed information corresponding to the abnormal number and send an interrupt signal to the system master control module.

[0011] The system master control module is configured to receive an interrupt signal and perform an emergency abnormal reset operation on an external device of the full-scene eutectic die bonder according to the interrupt signal and start a buzzer alarm.

[0012] Further, the full-scene abnormal signal acquisition module comprises a signal filter, an abnormal data acquisition card and an abnormal signal judgment module connected in sequence, wherein:

[0013] The signal filter is configured to filter the collected device state signal.

[0014] The abnormal data acquisition card is configured to collect the abnormal signal and send it to the abnormal signal judgment module.

[0015] The abnormal signal judgment module is configured to obtain an ID number and an abnormal number of an abnormal device corresponding to the abnormal signal.

[0016] Further, the abnormal signal judgment module comprises an abnormal signal receiving module, a eutectic table fault knowledge graph construction module and a retrieval module, wherein:

[0017] The abnormal signal receiving module is configured to receive the abnormal signal sent by the abnormal data acquisition card.

[0018] The eutectic table fault knowledge graph construction module is configured to construct a eutectic table fault knowledge graph.

[0019] The retrieval module is configured to obtain the ID number and the abnormal number of the abnormal device corresponding to the abnormal signal according to the eutectic table fault knowledge graph construction module.

[0020] Further, the eutectic table fault knowledge graph construction module comprises a directed label graph construction module, a coreference resolution and entity disambiguation module, a loss function design module and a neural network training module connected in sequence, wherein:

[0021] The directed label graph construction module is configured to construct a directed label graph G=(V,E,R), wherein V, E and R represent a node set, an edge set and a relationship attribute set, respectively.

[0022] The coreference resolution and entity disambiguation module is configured to perform coreference resolution and entity disambiguation on the nodes.

[0023] The loss function design module is configured to design a loss function, and the specific formula of the loss function is.

[0024]

[0025] wherein, is the loss function, ω is the number of negative samples, T is the set of all negative samples and positive samples, For a subset of the edge set E of the directed labeled graph G, the relationship between nodes is a triple (s, r, o), wherein s is the subject node, o is the object node, r is the link relationship between the subject node s and the object node o, y represents a flag bit, y=1 for a positive sample and y=0 for a negative sample, f(s, r, o) represents a scoring function of the triple, and sigma(·) is a sigmoid activation function.

[0026] The neural network training module is configured to train the neural network according to the loss function The neural network is trained by using the gradient descent method, and when the loss function When the loss function is stable, for all positive and negative samples (s, r, o, y) in T, the scoring function f(s, r, o) is calculated, if sigma(f(s, r, o))>0.8, it is considered that there is strong correlation between the corresponding subject node s and the object node o, if the sample belongs to a positive sample, because the link connection already exists, no processing is performed; if it is a negative sample, a link relationship r is added between the subject node s and the object node o, so that the relationship between the nodes of the knowledge graph is updated and completed.

[0027] Further, the nodes in the directed labeled graph construction module are specifically one or more combinations of a telecommunication signal receiving GPIO port signal type, a sensor signal type, a eutectic table device ID signal type and a eutectic table device abnormal number signal type.

[0028] Further, the abnormal alarm master control module comprises a concurrent type intra-memory addressing module, an ASMA abnormal number matching module, an interrupt signal sending module and a display sending module connected in sequence, wherein:

[0029] The concurrent type intra-memory addressing module is configured to determine an abnormal set corresponding to the ID number of the abnormal device according to the ID number of the abnormal device.

[0030] The ASMA abnormal number matching module is configured to use an ASMA full-character fast matching algorithm and run in multiple threads in parallel to determine abnormal detailed information corresponding to the abnormal number from the abnormal set based on the abnormal set corresponding to the ID number of the abnormal device.

[0031] The interrupt signal sending module is configured to send an interrupt signal to the system master control module according to the ID number of the abnormal device and the abnormal detailed information corresponding to the abnormal number.

[0032] The display sending module is configured to display the ID number of the abnormal device and the abnormal detailed information corresponding to the abnormal number on a monitoring display, and write the ID number of the abnormal device and the abnormal detailed information corresponding to the abnormal number into a system error log.

[0033] The full-scene operation abnormal intelligent alarm method of the eutectic patch machine provided by the application comprises:

[0034] An external device signal is collected, and the external device signal is specifically a device state signal of an external device running in a full scene of a die bonding machine.

[0035] According to the external device signal, an abnormal signal is obtained.

[0036] The ID number and the abnormal number of the abnormal device corresponding to the abnormal signal are obtained.

[0037] According to the ID number and the abnormal number of the abnormal device, an alarm is given for the die bonding machine running in a full scene.

[0038] Further, obtaining the ID number and the abnormal number of the abnormal device corresponding to the abnormal signal comprises:

[0039] A die table fault knowledge graph is constructed.

[0040] Based on the die table fault knowledge graph, the abnormal signal is searched to obtain the ID number and the abnormal number of the abnormal device corresponding to the abnormal signal.

[0041] Further, according to the ID number and the abnormal number of the abnormal device, an alarm is given for the die bonding machine running in a full scene, which comprises:

[0042] According to the ID number of the abnormal device, a concurrent in-memory addressing algorithm is used to determine an abnormal set corresponding to the ID number of the abnormal device.

[0043] Based on the abnormal set corresponding to the ID number of the abnormal device, an ASMA full-character fast matching algorithm is used for multi-thread parallel running to match and determine abnormal detailed information corresponding to the abnormal number from the abnormal set.

[0044] According to the ID number of the abnormal device and the abnormal detailed information corresponding to the abnormal number, an alarm is given for the die bonding machine running in a full scene.

[0045] Further, based on the abnormal set corresponding to the ID number of the abnormal device, an ASMA full-character fast matching algorithm is used for multi-thread parallel running to match and determine abnormal detailed information corresponding to the abnormal number from the abnormal set, which comprises:

[0046] Step 1: define the abnormal number to be matched as a first string strO, and initialize the conversion result numO of the first string to be equal to 0, and let i = 1;

[0047] Step 2: update the numerical conversion result of the first string strO, and the update formula is specifically:

[0048] numO = numO + Ψ(strO[i]) × 10 3(i-1) ,

[0049] Where Ψ(·) represents the mapping function from string to number field, and strO[i] represents the i-th character of the first string strO;

[0050] Step 3: When 1≤i≤n, i=i+1, repeat step 2. When i>n, the ASMA conversion of the first string strO is completed, and the result is numO, where n represents the normalized length of the string.

[0051] Step 4: Let the string strS be the string representing the kth exception number in the exception set. k The numerical conversion result numS k Set the initial value to 0, and let j = 1;

[0052] Step 5: Update the string strS of the k-th exception ID in the exception set. k The numerical conversion result numS k The specific formula for updating is as follows:

[0053] numS k =numS k +Ψ(strS k [j])×10 3(j-1) ,

[0054] Where, strS k [j] represents the string strS, which represents the k-th exception ID in the exception set. k The j-th character;

[0055] Step 6: When 1 ≤ j ≤ n, j = j + 1, repeat step 5. When j > n, the string strS of the kth exception number in the exception set. k The ASMA conversion is complete, and the result is numS. k ;

[0056] Step 7: Combine the results from Step 3 and Step 6, numO and numS. k Perform a numerical comparison; if numO = numS k If numO ≠ numS, then proceed to Step 8; k If k = k + 1, then proceed to Step 5;

[0057] Step 8: Obtain the exception number string strS at this time. k And use key-value pairs to query and obtain detailed exception information corresponding to the exception number.

[0058] The beneficial effects of this invention specifically include:

[0059] (1), by constructing the eutectic station fault knowledge graph, get rid of the dependence on the host computer, make the full scene abnormal signal acquisition module can be independently and quickly from the collected electric signal judge the abnormal device id and abnormal number generated, lay the foundation for subsequent data processing.

[0060] (2), based on the link relationship completion algorithm of RKL-GCN, complete the relationship of each node in the eutectic station fault knowledge graph, improve the reasoning accuracy of the knowledge graph, realize the accurate comprehensive judgment of abnormal signal.

[0061] (3), using concurrent in-memory addressing algorithm and ASMA abnormal number fast matching method, accelerate the acquisition of abnormal detailed information, shorten the time from abnormal occurrence to abnormal determination as much as possible, reduce the damage of abnormality to each device in the eutectic chip mounter. BRIEF DESCRIPTION OF DRAWINGS

[0062] Figure 1 the structure block diagram of the full scene operation abnormal intelligent alarm device of the eutectic chip mounter of the second embodiment of the present application;

[0063] Figure 2 the structure block diagram of the full scene abnormal signal acquisition module of the second embodiment of the present application;

[0064] Figure 3 the structure block diagram of the abnormal alarm main control module of the second embodiment of the present application;

[0065] Figure 4 the flow chart of the full scene operation abnormal intelligent alarm method of the eutectic chip mounter of the third embodiment of the present application.

[0066] REFERENCE NUMERALS:

[0067] M1, external device; M2, full scene abnormal signal acquisition module; M3, abnormal alarm main control module; M4, system main control module. DETAILED DESCRIPTION

[0068] In order to facilitate the understanding of the present application, the following will be combined with the description and the preferred embodiments to make a more comprehensive and detailed description of the present application, but the protection scope of the present application is not limited to the following specific embodiments.

[0069] The embodiments of the present application will be described in detail below in combination with the drawings, but the present application can be implemented in various different ways limited and covered by the claims.

[0070] Embodiment one

[0071] The eutectic chip mounter full scene operation abnormal intelligent alarm device provided by the first embodiment of the present application comprises a full scene abnormal signal acquisition module, an abnormal alarm main control module and a system main control module connected with the external devices of the eutectic chip mounter full scene operation in turn, wherein:

[0072] The full-scene abnormal signal acquisition module is configured to acquire device state signals of the full-scene running external device of the eutectic die bonder, obtain abnormal signals and ID numbers and abnormal numbers of abnormal devices corresponding to the abnormal signals according to the device state signals.

[0073] The abnormal alarm master module is configured to obtain abnormal detailed information corresponding to the abnormal numbers and send an interrupt signal to the system master module.

[0074] The system master module is configured to receive the interrupt signal and perform an emergency abnormal reset operation on the full-scene running external device of the eutectic die bonder and start a buzzer alarm according to the interrupt signal.

[0075] The eutectic die bonder full-scene running abnormal intelligent alarm device provided by the embodiment of the application comprises a full-scene abnormal signal acquisition module, an abnormal alarm master module and a system master module connected with the full-scene running external device of the eutectic die bonder in sequence, wherein the full-scene abnormal signal acquisition module is configured to acquire device state signals of the full-scene running external device of the eutectic die bonder, obtain abnormal signals and ID numbers and abnormal numbers of abnormal devices corresponding to the abnormal signals according to the device state signals; the abnormal alarm master module is configured to obtain abnormal detailed information corresponding to the abnormal numbers and send an interrupt signal to the system master module; and the system master module is configured to receive the interrupt signal and perform an emergency abnormal reset operation on the full-scene running external device of the eutectic die bonder and start a buzzer alarm according to the interrupt signal. The technical problem that the prior art cannot accurately monitor and rapidly alarm the running abnormality of the eutectic die bonder is solved, the inference accuracy of the knowledge graph is improved by complementing the node relationship in the eutectic table fault knowledge graph, accurate comprehensive judgment of the abnormal signal is realized, the acquisition of the abnormal detailed information is accelerated by using the concurrent in-memory addressing algorithm and the ASMA abnormal number rapid matching method, the time from the abnormality to the determination of the abnormality is shortened as much as possible, and the damage of the abnormality to the devices in the eutectic die bonder is reduced.

[0076] Embodiment two

[0077] As Figure 1As shown, Embodiment 2 of the present invention provides an intelligent alarm device for all-scenario operation abnormalities of a eutectic surface mount machine. This device mainly consists of an external device M1, a full-scenario abnormal signal acquisition module M2, an abnormal alarm main control module M3, and a system main control module M4. The specific workflow is as follows: First, using detection sensors, voltage and current signals reflecting the status of each device are collected. These signals are then converted from analog signals to digital signals that can be processed by a computer using an AD converter. When an abnormality occurs, the full-scenario abnormal signal acquisition module determines the device ID and abnormality number corresponding to the abnormal signal. Subsequently, the abnormal alarm main control module quickly obtains the corresponding device name and detailed abnormality information, sends an interrupt signal to the system main control module, and simultaneously sends the device name and detailed abnormality information to the monitoring display, promptly alerting on-site personnel for troubleshooting and other processing. Upon receiving the interrupt signal, the system main control module performs an emergency abnormality reset operation on each device and activates a buzzer alarm to reduce the danger and loss caused by the abnormality. The following is a detailed description of each part of the system:

[0078] External device M1

[0079] The external device M1 of this invention mainly includes a nozzle motor, a robotic arm motor, an industrial camera, and a eutectic heating stage. To accurately obtain the specific operating status of each device, sensors such as infrared temperature sensors, Hall effect sensors, and voltage and current detectors are equipped according to the characteristics of the devices. This enables comprehensive monitoring of the operating status of each device and ensures that when a device malfunctions, the sensors quickly send an abnormal signal to the full-scene abnormal signal acquisition module.

[0080] M2 Full-Scene Abnormal Signal Acquisition Module

[0081] The structure of the full-scene abnormal signal acquisition module in this embodiment is as follows: Figure 2 As shown, it consists of a signal filter, an NI USB-6210 anomaly data acquisition card, and an MSP430 anomaly signal judgment module. First, external sensors acquire electrical signals from each device. Then, these signals are passed through a signal filter to eliminate signal fluctuations caused by environmental factors, preventing misjudgments of signal anomalies due to interference. Next, the NI USB-6210 anomaly data acquisition card acquires anomaly signals and sends them to the MSP430 anomaly signal judgment module. The MSP430 anomaly signal judgment module calculates the ID number and anomaly number of the anomaly device corresponding to the anomaly signal.

[0082] The MSP430 abnormal signal judgment module contains a reflow station fault knowledge graph composed of multiple types of nodes such as a telecommunication signal receiving GPIO port class, a sensor class, a reflow station device id class, and a reflow station device abnormal number class. The voltage and current of the GPIO port are collected and received, and the internal relationship between the nodes is used to jump between the nodes, and finally the device generating the abnormal signal is judged and the specific device id number and abnormal number information are outputted, realizing intelligent retrieval and providing basis for subsequent processes. The construction of the reflow station fault knowledge graph specifically includes the following steps. First, the nodes with the same meaning are classified into the same class through reference resolution. Then, entity disambiguation is performed on the nodes that may cause ambiguity, such as the word "module" which can refer to a software system function module or a hardware component module, so entity disambiguation is needed to confirm the actual meaning. Finally, the link relationship completion algorithm of RKL-GCN is used to fill in the relationship between the nodes of the knowledge graph to improve the accuracy of abnormality judgment. The link relationship completion algorithm of RKL-GCN is described as follows.

[0083] Let the reflow station fault knowledge graph be a directed label graph G=(V,E,R), where the node v i ∈V, the edge (v i ,r,v j )∈E, and the relationship attribute r∈R. The feature vector of the i-th node in the l+1-th layer can be obtained by the following formula,

[0084]

[0085] where is the feature vector of the i-th node in the l-th layer, σ is the sigmoid activation function, denotes the peripheral node in the relationship r with the node i, is a normalization constant, are the self-weight matrix and the relationship weight matrix, respectively.

[0086] Since the number of data sets is always limited, the relationships between nodes in the knowledge graph established by the fault data set are not comprehensive, so it is necessary to complete the relationships between nodes to improve the accuracy of knowledge graph reasoning. Let the edge E subset of the directed label graph G be Let the relationship between nodes be a triple (s,r,o), where s is the subject node and o is the object node. Let the L-th layer obtained by formula (1) be Define the scoring function f(s,r,o) of the triple, as shown in formula (2),

[0087] ​

[0088] R r is a diagonal matrix of the relation r.

[0089] The subject s or the object o of the corrupted true triple (positive sample) (s, r, o) forms a negative sample. A loss function is designed as follows,

[0090]

[0091] where ω is the number of negative samples, T is the set of all negative samples and positive samples, y is a flag, y = 1 for a positive sample and y = 0 for a negative sample.

[0092] According to the loss function , the neural network RKL-GCN is trained by using the gradient descent method, and when the loss function is stable, the score function f(s, r, o) of all positive and negative samples (s, r, o, y) ∈ T is calculated, and if σ(f(s, r, o)) > 0.8, it is considered that there is strong correlation between the subject node s and the object node o, and if the sample belongs to a positive sample, because the link connection already exists, it is not processed; if it is a negative sample, a link relationship r is added between the subject node s and the object node o, realizing the updating and completion of the relationship between each node of the knowledge graph.

[0093] Abnormal alarm master control module M3

[0094] The abnormal alarm master control module of the embodiment is as shown in Figure 3 , mainly including a concurrent type in-memory addressing algorithm, an ASMA abnormal number quick matching part, etc. First, after obtaining the device id number by the full-scene abnormal signal acquisition module, the concurrent type in-memory addressing algorithm is used to determine the abnormal set corresponding to the abnormal device, and the abnormal set of each device is composed of the corresponding device manual abnormality and the on-site experience abnormality of the device. Secondly, based on the abnormal set corresponding to the abnormal device obtained above, combined with the abnormal number obtained by the full-scene abnormal signal acquisition module, based on the ASMA full-character quick matching algorithm, multi-thread parallel running is performed to determine the abnormal detailed information corresponding to the abnormal number from the device abnormal set, which is as follows.

[0095] Take the abnormal number to be matched (i.e. the abnormal number obtained by the full-scene abnormal signal acquisition module) as a string strO, and define the abnormal number in the abnormal set corresponding to the abnormal device as a string strS k, k represents the kth abnormal number string in the abnormal data set, 1≤k≤N, N is the total number of abnormalities in the abnormal data set corresponding to the abnormal device. Due to the same management requirements of industrial equipment, the abnormal number format is uniform, so the length of the abnormal number string to be matched and all strings in the abnormal number set can be unified as n. Define strO[i] represents the ith character in the string strO, 1≤i≤n, define strS k [j] represents the jth character in the string strS k , 1≤j≤n. The numerical results of strO, strS k converted by ASMA are represented by numO, numS k respectively, and represents a mapping from a string to a numerical domain, and the mapping rule is ASCII. The flow of the ASMA full-character fast matching algorithm is summarized as follows.

[0096] Step1: Let the numerical conversion result numO of the abnormal number string strO to be matched be initialized to 0, and prepare for subsequent saving of ASMA conversion results. Let i=1, and prepare for ASMA conversion.

[0097] Step2: Update the numerical conversion result numO of the abnormal number string strO to be matched by the following formula.

[0098] numO=numO+Ψ(strO[i])×10 3(i-1) (4)

[0099] Step3: When 1≤i≤n, i=i+1, repeat step2; when i>n, the ASMA conversion of the abnormal number to be matched is completed, and the result is numO.

[0100] Step4: Let the numerical conversion result numO of the kth abnormal number string strS k in the abnormal data set be initialized to 0, and let j=1, and prepare for ASMA conversion.

[0101] Step5: Update the numerical conversion result numS k of the kth abnormal number string strS k in the abnormal data set by the following formula.

[0102] numS k =numS k +Ψ(strS k [j])×10 3(j-1) (5)

[0103] Step 6: When 1 ≤ j ≤ n, j = j + 1, repeat step 5; when i > n, the ASMA transformation of the exception number to be matched is completed, and the result is numS. k .

[0104] Step 7: Combine the results from Step 3 and Step 6, numO and numS. k Directly compare the values; if numO = numS k If numO ≠ numS, then proceed to Step 8; k If k = k + 1, then proceed to Step 5.

[0105] Step 8: Obtain the exception number strS at this time. k This indicates a successful match.

[0106] The exception number, determined using the above method, is then used to retrieve detailed exception information corresponding to that number via key-value pair lookup. Finally, the exception information is output, written to the system error log, displayed on the monitoring monitor, with the exception flag set to 1, and an interrupt signal sent to the system main control module.

[0107] System main control module M4

[0108] System main control module such as Figure 1 As shown, after receiving the interrupt signal sent by the abnormal alarm main control module M3, the system performs power-off or reset operations on the relevant equipment according to the abnormal equipment and specific abnormal information, such as stopping the heating of the eutectic stage and resetting the robotic arm. At the same time, it sounds an alarm through a buzzer to promptly notify on-site personnel to check the equipment and troubleshoot the fault.

[0109] The purpose of this invention is to design an intelligent alarm system for abnormal operation of a eutectic bonding machine across all scenarios, solving the problem of monitoring and rapidly alarming abnormal operation of the eutectic bonding machine in all scenarios. The invention also aims to design an abnormal signal acquisition module for all scenarios, utilizing filters, a data acquisition card, and an MSP430 chip integrating a eutectic bonding station fault knowledge graph to convert abnormal electrical signals into specific abnormal device IDs and abnormal numbers. Furthermore, the invention designs an abnormal alarm main control module that achieves rapid search for abnormal devices through a concurrent memory-like addressing algorithm, and simultaneously achieves rapid matching of abnormal numbers using ASMA abnormal numbering through multi-threaded parallelism, solving the problem of rapid querying and obtaining abnormal information.

[0110] Example 3

[0111] The following is in conjunction with the appendix Figure 4 The intelligent method for handling all-scenario operational anomalies in the eutectic bonding machine according to the present invention will be described in further detail. After all modules are connected, operation begins:

[0112] 1. Through the external device sensor, the running information of the device is collected continuously.

[0113] 2. The collected device running information is converted into digital electrical signals by an AD converter.

[0114] 3. The input full-scene abnormal signal collection module reads the digital electrical signals, and judges whether there is an abnormality by using the eutectic table fault knowledge graph.

[0115] 4. If there is no abnormality, return to step 1 to continue signal collection.

[0116] 5. If there is an abnormality, the eutectic table fault knowledge graph outputs the abnormal device ID number and the abnormal number.

[0117] 6. Based on the abnormal device ID number and the abnormal number, the memory addressing in the abnormal alarm master module and the ASMA joint multi-thread fast retrieval algorithm are used to efficiently and quickly obtain the matched abnormal device name and abnormal detailed information.

[0118] 7. The abnormal device name and the abnormal detailed information are output to a display, so that the on-site staff can view the abnormal information on the monitoring display.

[0119] 8. The abnormal flag in the abnormal alarm master module is set to 1, and an interrupt signal is sent to the system master module.

[0120] 9. The system master module receives the interrupt signal, controls the device that has an abnormality to perform reset, breakpoint and other operations, simultaneously starts a buzzer alarm, and uses sound to timely remind the on-site staff.

[0121] The full-scene running abnormality intelligent alarm method of the eutectic chip mounter provided by the embodiment of the application collects external device signals, obtains abnormal signals according to the external device signals, obtains the ID number and the abnormal number of the abnormal device corresponding to the abnormal signals, and alarms the full-scene running abnormality of the eutectic chip mounter according to the ID number and the abnormal number of the abnormal device, thereby solving the technical problem that the prior art cannot accurately monitor and quickly alarm the running abnormality of the eutectic chip mounter. By completing the relationship between each node in the eutectic table fault knowledge graph, the reasoning accuracy of the knowledge graph is improved, accurate comprehensive judgment of the abnormal signal is realized, the concurrent memory addressing algorithm and the ASMA abnormal number fast matching method are used to accelerate the acquisition of abnormal detailed information, the time from abnormality occurrence to abnormality determination is shortened as much as possible, and the damage of the abnormality to each device in the eutectic chip mounter is reduced.

[0122] The above merely describes the preferred embodiments of the present application, and is not used to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A smart alarm device for abnormal operation of a eutectic bonding machine in all scenarios, characterized in that, The device includes a full-scene abnormal signal acquisition module, an abnormal alarm main control module, and a system main control module, which are sequentially connected to the external equipment of the eutectic bonding machine for full-scene operation. The full-scene abnormal signal acquisition module is used to acquire device status signals of external devices operating in the entire scenario of the eutectic bonding machine, obtain abnormal signals based on the device status signals, and acquire the ID number and abnormal number of the abnormal device corresponding to the abnormal signal. The full-scene abnormal signal acquisition module includes a signal filter, an abnormal data acquisition card, and an abnormal signal judgment module connected in sequence. The signal filter is used to filter the acquired device status signal; The abnormal data acquisition card is used to acquire abnormal signals and send them to the abnormal signal judgment module; The abnormal signal judgment module is used to obtain the ID number and abnormal number of the abnormal device corresponding to the abnormal signal. The abnormal signal judgment module includes an abnormal signal receiving module, a eutectic stage fault knowledge graph construction module, and a retrieval module, wherein: The abnormal signal receiving module is used to receive abnormal signals sent by the abnormal data acquisition card; The eutectic stage fault knowledge graph construction module is used to construct a eutectic stage fault knowledge graph. This module includes, in sequence, a directed label graph construction module, an affixation resolution and entity disambiguation module, a loss function design module, and a neural network training module, wherein: The directed label graph construction module is used to construct a directed label graph G = (V, E, R), where V, E, and R represent the set of nodes, the set of edges, and the set of relation attributes, respectively. The substituent resolution and entity disambiguation modules are used to resolve substituents and disambiguate entities for nodes. The loss function design module is used to design the loss function, the specific formula of which is: in, Let be the loss function, ω be the number of negative samples, and T be the set of all negative and positive samples. Let G be a subset of the edge set E of the directed label graph G. Let the relationship between nodes be a triple (s,r,o), where s is the subject node, o is the object node, r is the link relationship between subject node s and object node o, y represents the flag bit, y=1 for positive samples and y=0 for negative samples, f(s,r,o) represents the scoring function of the triple, and σ(·) is the sigmoid activation function. The neural network training module is used to train the neural network based on the loss function. When training a neural network using gradient descent, the loss function... When stable, for all positive and negative samples (s,r,o,y)∈T, the scoring function f(s,r,o) is calculated. If σ(f(s,r,o))>0.8, it is considered that there is a strong correlation between the corresponding subject node s and object node o. If the sample is a positive sample, no processing is done because the link connection already exists. If it is a negative sample, a link relationship r is added between the subject node s and the object node o to update and complete the relationship between the nodes of the knowledge graph. The retrieval module is used to obtain the ID number and abnormal number of the abnormal device corresponding to the abnormal signal based on the eutectic stage fault knowledge graph construction module. The abnormal alarm main control module is used to obtain detailed abnormal information corresponding to the abnormal number and send an interrupt signal to the system main control module; The system's main control module is used to receive interrupt signals and, based on the interrupt signals, perform emergency abnormal reset operations on external devices operating in all scenarios of the eutectic bonding machine and activate the buzzer alarm.

2. The intelligent alarm device for abnormal operation of the eutectic bonding machine in all scenarios according to claim 1, characterized in that, The nodes in the directed label graph construction module are specifically composed of one or more combinations of the following: electrical signal receiving GPIO port signal type, sensor signal type, eutectic device ID signal type, and eutectic device fault number signal type.

3. The intelligent alarm device for abnormal operation of a eutectic bonding machine in all scenarios according to any one of claims 1-2, characterized in that, The abnormal alarm main control module includes a concurrent memory addressing module, an ASMA abnormal number matching module, an interrupt signal sending module, and a display module connected in sequence, wherein: The concurrent memory addressing module is used to determine the exception set corresponding to the ID number of the exception device based on the ID number of the exception device; The ASMA exception number matching module is used to determine the detailed exception information corresponding to the exception number by matching the exception set corresponding to the ID number of the exception device, using the ASMA full character fast matching algorithm, running in parallel with multiple threads, and matching from the exception set. The interrupt signal sending module is used to send an interrupt signal to the system main control module according to the ID number of the abnormal device and the abnormal details corresponding to the abnormal number; The display module is used to display the ID number of the abnormal device and the detailed abnormal information corresponding to the abnormal number on the monitoring display, and to write the ID number of the abnormal device and the detailed abnormal information corresponding to the abnormal number into the system error log.

4. A method for providing intelligent alarm for abnormal operation of a eutectic chip mounter in all scenarios using the intelligent alarm device for abnormal operation of a eutectic chip mounter according to any one of claims 1-3, comprising: Collect external device signals, specifically the device status signals of external devices operating in all scenarios of the eutectic bonding machine; Obtain abnormal signals based on signals from external devices; Obtain the ID number and anomaly number of the abnormal device corresponding to the abnormal signal; Based on the ID number and anomaly number of the abnormal device, alarms are triggered for abnormal operation of the eutectic chip mounter across all scenarios.

5. The intelligent alarm method for abnormal operation of a eutectic bonding machine in all scenarios according to claim 4, characterized in that, The ID number and anomaly number of the abnormal device corresponding to the abnormal signal are obtained as follows: Constructing a knowledge graph of eutectic stage faults; Based on the knowledge graph of eutectic stage faults, abnormal signals are retrieved to obtain the ID number and abnormal number of the abnormal device corresponding to the abnormal signal.

6. The intelligent alarm method for abnormal operation of a eutectic bonding machine in all scenarios according to claim 5, characterized in that, Based on the ID number and anomaly number of the faulty device, alarms are triggered for all scenarios of abnormal operation of the eutectic bonding machine, including: Based on the ID number of the abnormal device, a concurrent class memory addressing algorithm is used to determine the set of abnormalities corresponding to the ID number of the abnormal device. Based on the exception set corresponding to the ID number of the abnormal device, the ASMA full character fast matching algorithm is used and multi-threaded parallel operation is performed to match and determine the detailed exception information corresponding to the exception number from the exception set. Based on the ID number of the abnormal device and the detailed abnormal information corresponding to the abnormal number, alarms are triggered for abnormal operation of the eutectic chip mounter in all scenarios.

7. The intelligent alarm method for abnormal operation of a eutectic bonding machine in all scenarios according to claim 6, characterized in that, Based on the exception set corresponding to the ID number of the abnormal device, the ASMA full-character fast matching algorithm is used, and multi-threaded parallel operation is performed to match and determine the detailed exception information corresponding to the exception number from the exception set, including: Step 1: Define the exception number to be matched as the first string strO, and initialize the conversion result numO of the first string to 0, and let i = 1; Step 2: Update the numerical conversion result of the first string strO. The specific update formula is as follows: numO=numO+Ψ(strO[i])×10 3(i-1) , Where Ψ(·) represents the mapping function from string to number field, and strO[i] represents the i-th character of the first string strO; Step 3: When 1≤i≤n, i=i+1, repeat step 2. When i>n, the ASMA conversion of the first string strO is completed, and the result is numO, where n represents the normalized length of the string. Step 4: Let the string strS be the string representing the kth exception number in the exception set. k The numerical conversion result numS k Set the initial value to 0, and let j = 1; Step 5: Update the string strS of the kth exception ID in the exception set. k The numerical conversion result numS k The specific formula for updating is as follows: numberS k =numberS k +Ψ(strS k [j])×10 3(j-1) , Where, strS k [j] represents the string strS, which represents the k-th exception ID in the exception set. k The j-th character; Step 6: When 1 ≤ j ≤ n, j = j + 1, repeat step 5. When j > n, the string strS of the kth exception number in the exception set. k The ASMA conversion is complete, and the result is numS. k ; Step 7: Combine the results from Step 3 and Step 6, numO and numS. k Perform a numerical comparison; if numO = numS k If numO ≠ numS, then proceed to Step 8; k If k = k + 1, then proceed to Step 5; Step 8: Obtain the exception number string strS at this time. k And use key-value pairs to query and obtain detailed exception information corresponding to the exception number.

Images