CN117808052B - Vacuum environment-based mechanical arm load self-adaption method and system - Google Patents
Vacuum environment-based mechanical arm load self-adaption method and system Download PDFInfo
- Publication number
- CN117808052B CN117808052B CN202410235265.3A CN202410235265A CN117808052B CN 117808052 B CN117808052 B CN 117808052B CN 202410235265 A CN202410235265 A CN 202410235265A CN 117808052 B CN117808052 B CN 117808052B
- Authority
- CN
- China
- Prior art keywords
- load
- data
- torque
- value
- correction
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 93
- 238000012937 correction Methods 0.000 claims abstract description 128
- 238000011156 evaluation Methods 0.000 claims abstract description 76
- 230000002159 abnormal effect Effects 0.000 claims abstract description 73
- 230000008569 process Effects 0.000 claims abstract description 24
- 238000012544 monitoring process Methods 0.000 claims abstract description 21
- 238000012360 testing method Methods 0.000 claims description 89
- 238000012549 training Methods 0.000 claims description 39
- 230000015654 memory Effects 0.000 claims description 16
- 238000010801 machine learning Methods 0.000 claims description 15
- 230000006978 adaptation Effects 0.000 claims description 12
- 238000003860 storage Methods 0.000 claims description 11
- 238000004364 calculation method Methods 0.000 claims description 10
- 238000004590 computer program Methods 0.000 claims description 10
- 238000004422 calculation algorithm Methods 0.000 claims description 8
- 238000013528 artificial neural network Methods 0.000 claims description 7
- 238000013527 convolutional neural network Methods 0.000 claims description 6
- 125000004122 cyclic group Chemical group 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 6
- 238000012795 verification Methods 0.000 claims description 6
- 230000000694 effects Effects 0.000 claims description 4
- 230000008859 change Effects 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 3
- 238000004458 analytical method Methods 0.000 abstract description 5
- 230000007246 mechanism Effects 0.000 abstract description 5
- 235000012431 wafers Nutrition 0.000 description 34
- 239000004065 semiconductor Substances 0.000 description 6
- 230000003044 adaptive effect Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 230000001133 acceleration Effects 0.000 description 3
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000005259 measurement Methods 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000005484 gravity Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 230000005856 abnormality Effects 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000009472 formulation Methods 0.000 description 1
- 230000007787 long-term memory Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Abstract
The invention discloses a vacuum environment-based mechanical arm load self-adaption method and a system, and particularly relates to the technical field of wafer handling, wherein the method comprises the steps of collecting load monitoring data of a mechanical arm in a wafer handling process at a moment T, acquiring a historical load evaluation data set, and acquiring a load evaluation coefficient of a load sensor at a moment T+a in the future according to a pre-constructed load learning model, wherein the load monitoring data comprises actual load data and first target load data; determining whether the load sensor at the time of T+a is in an abnormal load state according to the predicted load evaluation coefficient; according to the invention, the abnormal load judging mechanism is loaded in the mechanical arm, the analysis of the abnormal load state is carried out when the mechanical arm conveys the wafer, the torque of the actuator is corrected, the load data is adaptively adjusted based on the torque correction value of the actuator, the mechanical arm is prevented from being abnormal in the wafer conveying process, and the safety of the mechanical arm for conveying the wafer is improved.
Description
Technical Field
The invention relates to the technical field of wafer handling, in particular to a vacuum environment-based mechanical arm load self-adaption method and system.
Background
In a vacuum environment, the semiconductor mechanical arm is used as an executing mechanism in the semiconductor conveying process, and the reliability of the semiconductor mechanical arm is related to the operation performance and the possibility level of the whole conveying system, so that the semiconductor mechanical arm has important significance in abnormal load monitoring; the operation process of the mechanical arm mainly relates to components such as a motor, a sensor, an actuator and the like, and in the mechanical arm, the accuracy of a load sensor is critical to the weight measurement of a carrying load. After the load sensor is replaced, calibration is usually required to ensure the accuracy; if calibration is not performed, the carrying load value cannot be accurately measured, so that the load error is increased, and further the mechanical arm system cannot obtain reliable load information, so that the wafer is damaged due to the load information error.
At present, the zero point and the sensitivity of a load sensor are mainly adjusted through temperature calibration and static state in the prior art so as to ensure the accuracy of the load sensor; or the existing method of load self-adaptation of the mechanical arm mainly compensates around the gravity of the mechanical arm and the gravity of the load driven by the mechanical arm, so that the mechanical arm stably moves towards a target position, for example, chinese patent publication No. CN113319844a discloses a mechanical arm control method, control equipment and a robot, but the inventor researches and uses the above method and the prior art find that the above method and the prior art have at least the following partial defects:
(1) Only through a single load sensor, a larger load error still exists;
(2) The lack of a load evaluation mechanism cannot predict the abnormality of the load sensor in the mechanical arm in advance, and therefore the mechanical arm carries the wafer based on abnormal load data, resulting in unstable wafer carrying.
Therefore, the invention provides a vacuum environment-based mechanical arm load self-adaption method and a vacuum environment-based mechanical arm load self-adaption system.
Disclosure of Invention
In order to overcome the above-mentioned drawbacks of the prior art, the present invention provides a load adaptive method and system for a mechanical arm based on a vacuum environment, so as to solve the above-mentioned problems in the prior art.
In order to achieve the above purpose, the present invention provides the following technical solutions: the vacuum environment-based mechanical arm load self-adaption method comprises the following steps:
collecting load monitoring data of the mechanical arm in the process of carrying the wafer at the moment T, acquiring a historical load evaluation data set, and acquiring a load evaluation coefficient of a load sensor at the moment T+a in the future according to a pre-constructed load learning model, wherein the load monitoring data comprises actual load data and first target load data; a is an integer greater than zero;
Determining whether the load sensor at the moment T+a is in an abnormal load state according to the predicted load evaluation coefficient, if not, continuing to convey the wafer by the mechanical arm under the first target load data, and enabling T=T+a+b, and if so, acquiring load characteristic data of the load sensor at the moment T+a; b is an integer greater than zero;
When the load sensor is in an abnormal load state, recording a torque value of the actuator, correcting the torque value of the actuator according to load characteristic data and a pre-constructed correction model, and determining actual load data of the mechanical arm based on the corrected torque value of the actuator; the load characteristic data comprises current of an actuator and torque of a motor;
And calculating a load difference value between the first target load data and the actual load data, judging whether abnormal handling exists, inputting the first target load data and the actual load data into a pre-constructed load correction model according to the abnormal handling to obtain second target load data, and carrying out wafer handling by the mechanical arm according to the second target load data.
Further, obtaining the load evaluation coefficient of the load sensor at the future time t+a according to the pre-constructed load learning model includes:
Acquiring voltage difference data of a load sensor, temperature difference data of the load sensor and vibration difference data of the load sensor from the moment T-b to the moment T;
The method comprises the steps of inputting voltage difference data of a load sensor, temperature difference data of the load sensor and vibration difference data of the load sensor from time T-b to time T into a pre-built load learning model to obtain a load evaluation coefficient of the load sensor at time T+a;
Acquiring a historical load evaluation data set, wherein the historical load evaluation data set comprises voltage difference data of a load sensor, temperature difference data of the load sensor, vibration difference data of the load sensor and corresponding load evaluation coefficients in n groups of time intervals, and n is an integer larger than zero;
Dividing a historical load evaluation data set for training a load learning model into a data training set and a data testing set, constructing a first regression model, taking voltage difference data of a load sensor, temperature difference data of the load sensor and vibration difference data of the load sensor in each group of preset time intervals as inputs of the first regression model, taking a load evaluation coefficient as output of the first regression model, and training the first regression model to obtain an initial load learning model; and evaluating the model effect of the initial load learning model by using a mean square error algorithm, and screening the corresponding initial load learning model with the evaluation value larger than or equal to the preset evaluation value as a load learning model.
Further, the method for obtaining the load evaluation coefficient comprises the following steps:
based on the mechanical arm in a constant-speed rotation stage, acquiring a load value of a load sensor in a preset time interval; obtaining a load standard value of the load sensor in a preset time interval; the load standard value is an average load value in a plurality of preset time intervals;
extracting a load value of the load sensor in a preset time interval at each moment, and carrying out formulated calculation on the load value and a load standard value to obtain a load evaluation coefficient of the load sensor; the calculation formula of the load evaluation coefficient is as follows:
；
Wherein: representing load assessment factor,/> Representing a load value at the e-th time; /(I)And (5) representing a load standard value, wherein E is the total time in a preset time interval.
Further, the method for determining whether the load sensor at the time of t+a is in an abnormal load state includes:
comparing the load evaluation coefficient with a preset coefficient gradient threshold value; the preset coefficient gradient threshold value comprises And/>，/>＞/>；
If it is＞/>Judging that the load sensor at the time of T+a is in an abnormal load state;
If it is Judging that the load sensor at the time of T+a is in a normal load state;
If it is And judging that the load sensor at the time T+a is in an abnormal load state.
Further, the method for correcting the torque value of the actuator according to the load characteristic data and the pre-constructed correction model comprises the following steps:
inputting the load characteristic data into a pre-constructed correction model to obtain a torque correction value;
Accumulating the torque correction value and the torque value to obtain a torque value after parameter correction;
The pre-constructed correction model is generated according to torque test data in a training way; the torque test data includes at least a relationship of a torque correction value to each ampere current of the actuator and a relationship of a torque correction value to each newton meter torque of the motor.
Further, the method for acquiring the torque test data comprises the following steps:
step a1: placing the test executor in a current change test environment, testing the torque value of the test executor, and placing the standard executor in a standard current test environment;
Step a2: in the test process, when the current of the test actuator is set to be p ampere, a first torque value of the actuator per newton meter is measured, and p is an integer larger than zero;
Step a3: under a standard current testing environment set by a standard actuator, acquiring a first standard torque value of each Newton meter of the standard actuator;
Step a4: taking the difference value between the first torque value and the first standard torque value as a first torque difference, comparing the first torque difference with a preset first torque error interval, if the first torque difference does not belong to the preset first torque error interval, enabling p=p+1, and returning to the step a2; if the first torque difference belongs to a preset first torque error interval, taking the first torque difference as a torque correction value, and binding and correlating a p-th ampere with the torque correction value to obtain the relation between the torque correction value and the p-th ampere current of the actuator;
Step a5: repeating the steps a2 to a4 until P is equal to the set current and the cycle is ended, obtaining the relation between the torque correction value and each ampere current of the actuator, wherein P is an integer larger than zero;
The method for acquiring the torque test data further comprises the following steps:
step b1: placing the test executor in a motor torque variation environment, testing the torque of the executor, and placing the standard executor in a set standard motor torque test environment;
step b2: under a test environment, when the motor torque of the test actuator is set to be w Newton-Mi Zhuaiju, a second torque value of the actuator per Newton-meter is measured, and w is an integer larger than zero;
Step b3: under a set standard motor torque test environment, obtaining a second standard torque value of each Newton meter of the standard actuator;
Step b4: taking the difference value between the second torque value and the second standard torque value as a second torque difference, comparing the second torque difference with a preset second torque error interval, if the second torque difference does not belong to the preset second torque error interval, letting w=w+1, and returning to the step b2; if the second torque difference belongs to a preset second torque error section, taking the second torque difference as a torque correction value, and binding and associating the w < Newton >. Mi Niuju with the torque correction value of the motor to obtain the relation between the torque correction value of the actuator and the w < Newton >. Meter torque of the motor;
Step b5: repeating the steps b 2-b 4 until W equals to the set torque W, ending the cycle, and obtaining the relation between the torque correction value of the actuator and each Newton meter torque of the motor, wherein W is an integer larger than zero.
Further, the method for generating the pre-constructed correction model comprises the following steps:
Dividing the torque test data into a torque correction training set and a torque correction test set; constructing a machine learning model, taking the current of an actuator and the torque of a motor in a torque correction training set as input data of the machine learning model, taking the torque correction value in the torque correction training set as output data of the machine learning model, and training the machine learning model to obtain an initial correction model; performing model verification on the initial correction model by using the torque correction test set, and outputting the initial correction model meeting the preset prediction error as a pre-constructed correction model; the machine learning model is one of a cyclic neural network, a convolutional neural network or a long-short-term memory network;
and determining actual load data of the load sensor based on the torque value after parameter correction, wherein the calculation formula of the actual load data is as follows:
；
wherein, Weight factor representing torque correction value,/>The weight factor indicating the arm length of the robot arm, and C is a correction constant of actual load data.
Further, the method for determining whether abnormal handling exists comprises the steps of:
Calculating a load difference value between the first target load data and the actual load data;
Comparing the load difference value with a load difference value threshold value;
If the load difference value is greater than or equal to the load difference value threshold value, judging that abnormal transportation exists;
If the load difference is smaller than the load difference threshold, judging that abnormal conveying does not exist.
Further, the second target load data is generated based on load test data training; the load test data acquisition method comprises the following steps:
Step c1: obtaining the ith abnormal handling, wherein i is an integer greater than zero;
step c2: acquiring first target load data of load sensor according to ith abnormal handling And according to the first target load data/>Actual load data/>, of load sensor under control；
Step c3: acquiring first target load dataAnd actual load data/>Is a load error of (a);
Step c4: judging whether the load error is equal to zero, if the load error is not equal to zero, further judging whether the load error is greater than zero or less than zero, if the load error is less than zero, then performing data processing on the first target load Increment, let/>=/>+G, and returning to step c2; if greater than zero, for the first target load data/>Decrementing, let/>=/>-G and returning to step c2; if the load error is equal to zero, the first target load data/>, after load adjustment, is obtainedAs second target load data/>And/>, first target load dataAnd actual load data/>And second target load data/>Correlating to obtain a group of first target load data/>And actual load data/>And second target load data/>G is an integer greater than zero;
Step c5: repeating the steps c 2-c 4 until i=i, ending the cycle to obtain H groups of first target load data And actual load data/>And second target load data/>To H groups of first target load data/>And actual load data/>And second target load data/>As load test data, I is the total number of abnormal handling, I, H is an integer greater than zero.
Further, the generation logic of the pre-constructed load correction model is as follows:
Dividing load test data into a load training set and a load test set;
Constructing a network learning model, taking first target load data and actual load data in a load training set as input data of the network learning model, taking second target load data in the load training set as output data of the network learning model, and training the network learning model to obtain an initial load correction model;
performing model verification on the initial load correction model by using a load test set, and outputting the initial load correction model meeting the preset prediction error as a pre-constructed load correction model;
The network learning model is one of a convolutional neural network, a cyclic neural network or a long-short-term memory network.
In a second aspect, the present invention provides a vacuum environment-based mechanical arm load adaptive system, configured to implement the vacuum environment-based mechanical arm load adaptive method described above, including:
The load evaluation module is used for collecting load monitoring data of the mechanical arm in the process of carrying the wafer at the moment T, acquiring a historical load evaluation data set, and acquiring a load evaluation coefficient of a load sensor at the moment T+a in the future according to a pre-constructed load learning model, wherein the load monitoring data comprises actual load data and first target load data; a is an integer greater than zero;
The load judging module is used for determining whether the load sensor at the moment T+a is in an abnormal load state according to the predicted load evaluation coefficient, if not, the mechanical arm continues to carry the wafer under the first target load data, and causes T=T+a+b, and if so, the load characteristic data of the load sensor at the moment T+a is obtained; b is an integer greater than zero;
the torque correction module is used for recording the torque value of the actuator when the load sensor is in an abnormal load state, correcting the torque value of the actuator according to the load characteristic data and a pre-constructed correction model, and determining the actual load data of the mechanical arm based on the corrected torque value of the actuator; the load characteristic data comprises current of an actuator and torque of a motor;
the load self-adaptation module is used for calculating a load difference value between the first target load data and the actual load data, judging whether abnormal handling exists, inputting the first target load data and the actual load data into a pre-constructed load correction model according to the abnormal handling, obtaining second target load data, and carrying out wafer handling by the mechanical arm according to the second target load data.
In a third aspect, the present invention provides an electronic device, comprising: a processor and a memory, wherein the memory stores a computer program for the processor to call;
and the processor executes the load self-adaption method based on the vacuum environment mechanical arm by calling the computer program stored in the memory.
In a fourth aspect, the present invention provides a computer readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the vacuum environment based mechanical arm load adaptation method described above.
The invention has the technical effects and advantages that:
According to the invention, load monitoring data of the mechanical arm in the process of carrying the wafer at the moment T is collected, a historical load evaluation data set is obtained, and a load evaluation coefficient of a load sensor at the moment T+a in the future is obtained according to a pre-built load learning model, wherein the load monitoring data comprises actual load data and first target load data; determining whether the load sensor at the moment T+a is in an abnormal load state according to the predicted load evaluation coefficient, if not, continuing to convey the wafer by the mechanical arm under the first target load data, and enabling T=T+a+b, and if so, acquiring load characteristic data of the load sensor at the moment T+a; when the load sensor is in an abnormal load state, recording a torque value of the actuator, correcting the torque value of the actuator according to load characteristic data and a pre-constructed correction model, and determining actual load data of the mechanical arm based on the corrected torque value of the actuator; the load characteristic data comprises current of an actuator and torque of a motor; calculating a load difference value between the first target load data and the actual load data, judging whether abnormal handling exists, inputting the first target load data and the actual load data into a pre-constructed load correction model according to the abnormal handling to obtain second target load data, and carrying out wafer handling by the mechanical arm according to the second target load data; according to the invention, the abnormal load judging mechanism is loaded in the mechanical arm, the analysis of the abnormal load state is carried out when the mechanical arm conveys the wafer, the torque of the actuator is corrected, the load data is adaptively adjusted based on the torque correction value of the actuator, the mechanical arm is prevented from being abnormal in the wafer conveying process, and the safety of the mechanical arm for conveying the wafer is improved.
Drawings
FIG. 1 is a flow chart of the method of example 1;
FIG. 2 is a flow chart of a method of torque test data acquisition of example 1;
FIG. 3 is a flow chart of another method of torque test data acquisition of example 1;
FIG. 4 is a schematic diagram of the system of example 2;
Fig. 5 is a schematic diagram of an electronic device in embodiment 3;
fig. 6 is a schematic diagram of a computer-readable storage medium according to embodiment 4.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Furthermore, the drawings are merely schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and thus a repetitive description thereof will be omitted. Some of the block diagrams shown in the figures are functional entities and do not necessarily correspond to physically or logically separate entities. The functional entities may be implemented in software or in one or more hardware modules or integrated circuits or in different networks and/or processor methods and/or microcontroller methods.
It will be understood that, although the terms "first," "second," etc. may be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another element. For example, a first element could be termed a second element, and a similar second element could be termed a first element, without departing from the scope of example embodiments. The term "and/or" as used herein includes any and all combinations of one or more of the associated listed items.
Example 1
Referring to fig. 1, the disclosure of the present embodiment provides a vacuum environment-based mechanical arm load adaptive method, which includes:
S1, collecting load monitoring data of a mechanical arm in a wafer carrying process at a moment T, acquiring a historical load evaluation data set, and acquiring a load evaluation coefficient of a load sensor at a moment T+a in the future according to a pre-constructed load learning model, wherein the load monitoring data comprises actual load data and first target load data; a is an integer greater than zero;
It should be appreciated that: in the process of carrying the wafer, the load sensor obtains load monitoring data in five stages: an initial stage, an acceleration stage, a uniform rotation stage, a deceleration stage and a placement stage; in the initial stage, when the mechanical arm starts to carry the wafer, the reading of the load sensor has an initial value; in the acceleration stage, the mechanical arm starts to accelerate, and the initial value is correspondingly increased due to inertial force caused by acceleration; in the constant-speed rotation stage, the reading of the load sensor is relatively stable; in the deceleration stage, the mechanical arm decelerates or stops moving, so that the inertia force generated in the deceleration process and the reaction force generated when the mechanical arm stops are reduced, and the numerical value of the load sensor is reduced; in the placing stage, the mechanical arm releases the force for carrying the wafer, so that the reading of the load sensor is continuously reduced until the reading is 0; the first target load data includes, but is not limited to, a reading on a load sensor;
It should be noted that, the actuator of the mechanical arm is installed with a load sensor, the load sensor provides real-time load information of the actuator, the control system adjusts the action of the actuator according to the real-time load information, when the mechanical arm carries wafers with the same specification and is in a constant rotation stage, the load sensor is usually described to work repeatedly for a long time with a relatively stable load period, and as the running time of the load sensor is continuously increased, the obtained real-time load information is abnormal, so that the actuator faces complex working condition changes, if the real-time load information is not processed in time, huge production accidents are easily caused, therefore, the application is designed according to the situation;
In the implementation, the method for acquiring the load evaluation coefficient of the mechanical arm at the future time T+a according to the pre-constructed load learning model comprises the following steps:
Step 101, acquiring voltage difference data of a load sensor, temperature difference data of the load sensor and vibration difference data of the load sensor from the moment T-b to the moment T; b is an integer greater than zero;
It should be noted that, the value of a is determined by a time interval preset by the arm manager, and if the time T is, for example, 8:00, and the set time interval is 10 minutes, the value of a is 10, namely 8 is obtained: 00-8:10, voltage difference data of a load sensor of the mechanical arm, temperature difference data of the load sensor and vibration difference data of the load sensor;
Continuing to describe, the voltage difference data includes a plurality of voltage difference values, each voltage difference value is obtained by calculating a difference value between two adjacent time intervals, and if the set time interval is 10 minutes, the obtained voltage difference data is 8:00-8:10, and two adjacent time intervals are 1 minute, then 8: voltage value of 01 minus 8: voltage value of 00, will be 8: voltage value of 02 minus 8:01, will 8: voltage value of 03 minus 8:02, and so on, until decremented to 8:10, obtaining all load difference values, and obtaining temperature difference data and vibration difference data by the same method, wherein redundant description is omitted;
the instability of the voltage value easily causes a decrease in the measurement performance of the load sensor, because the fluctuation of the voltage affects the electronic components of the sensor, thereby causing a measurement error.
102, Inputting voltage difference data of a load sensor, temperature difference data of the load sensor and vibration difference data of the load sensor from the moment T-b to the moment T to a pre-built load learning model to obtain a load evaluation coefficient of the load sensor at the moment T+a;
specifically, the method for generating the pre-constructed load learning model comprises the following steps:
Step 201, acquiring a historical load evaluation data set, wherein the historical load evaluation data set comprises voltage difference data of a load sensor, temperature difference data of the load sensor, vibration difference data of the load sensor and corresponding load evaluation coefficients in n groups of time intervals, and n is an integer larger than zero;
It should be noted that, the historical load evaluation data set is obtained by monitoring and preprocessing various sensors in a vacuum environment and is pre-stored in a mechanical arm database, the various sensors include but are not limited to a voltage sensor, a vibration sensor, a temperature sensor and the like, and the preprocessing includes but is not limited to data noise removal, data cleaning or error data deletion;
step 202, dividing a historical load evaluation data set for training a load learning model into a data training set and a data testing set, constructing a first regression model, taking voltage difference data of a load sensor, temperature difference data of the load sensor and vibration difference data of the load sensor in each group of preset time intervals as inputs of the first regression model, taking a load evaluation coefficient as output of the first regression model, and training the first regression model to obtain an initial load learning model; evaluating model effects of the initial load learning model by means of a mean square error algorithm, and screening corresponding initial load learning models with the evaluation value larger than or equal to a preset evaluation value as load learning models;
The load learning model comprises, but is not limited to, a K-nearest neighbor regression algorithm model, a neural network algorithm model, a long-term and short-term memory network algorithm model and the like; the mean square error algorithm has the following calculation formula: wherein: /(I) Representing the evaluation value/>Representing a characteristic sample,/>Representing a data test set,/>Representing the true value,/>Representing predicted value/>Representing the number of test samples in the data test set;
further stated, the method for obtaining the load evaluation coefficient is as follows:
Based on the mechanical arm in a constant-speed rotation stage, acquiring a load value of a load sensor in a preset time interval; obtaining a load standard value of the load sensor in a preset time interval; the load standard value is an average load value in a plurality of preset time intervals.
Extracting a load value of the load sensor in a preset time interval at each moment, and carrying out formulated calculation on the load value and a load standard value to obtain a load evaluation coefficient of the load sensor; the calculation formula of the load evaluation coefficient is as follows:
；
Wherein: representing load assessment factor,/> Representing a load value at the e-th time; /(I)The load standard value is represented, and E is the total time in a preset time interval;
It should be noted that, according to the load value and the load standard value, the load evaluation coefficient is obtained by performing formulation calculation, and meanwhile, the voltage difference data of the load sensor, the temperature difference data of the load sensor and the vibration difference data of the load sensor are also obtained, so that the voltage difference data of each group of load sensors, the temperature difference data of the load sensor and the vibration difference data of the load sensor correspond to one load evaluation coefficient; in other words, the voltage difference data of each group of load sensors, the temperature difference data of the load sensors and the vibration difference data of the load sensors establish a relation with a load value and a load standard value through a load learning model, wherein the relation is obtained through a historical load evaluation data set and is used for analyzing the load conditions of the load sensors in different time periods and generating corresponding load evaluation coefficients;
s2, determining whether the load sensor at the moment T+a is in an abnormal load state according to the predicted load evaluation coefficient, if not, continuously carrying the wafer by the mechanical arm under the first target load data, and enabling T=T+a+b, and if so, acquiring load characteristic data of the load sensor at the moment T+a;
In an implementation, a method of determining whether a load sensor at a time t+a is in an abnormal load state includes:
comparing the load evaluation coefficient with a preset coefficient gradient threshold value; the preset coefficient gradient threshold value comprises And/>，/>＞/>；
If it is＞/>Judging that the load sensor at the time of T+a is in an abnormal load state;
If it is Judging that the load sensor at the time of T+a is in a normal load state;
If it is Judging that the load sensor at the time of T+a is in an abnormal load state;
S3, when the load sensor is in an abnormal load state, recording a torque value of the actuator, correcting the torque value of the actuator according to load characteristic data and a pre-constructed correction model, and determining actual load data of the mechanical arm based on the corrected torque value of the actuator;
in an implementation, the load signature data includes current of the actuator and torque of the motor;
in an implementation, performing parameter correction on a torque value of an actuator according to load characteristic data and a pre-constructed correction model, including:
inputting the load characteristic data into a pre-constructed correction model to obtain a torque correction value;
Accumulating the torque correction value and the torque value to obtain a torque value after parameter correction;
specifically, the pre-constructed correction model is generated according to torque test data in a training way; the torque test data at least comprises a relation between a torque correction value and each ampere current of an actuator and a relation between the torque correction value and each newton meter torque of a motor;
It should be noted that, in the wafer carrying process by the mechanical arm, when the load sensor has an abnormal load state, the torque value can be corrected by changing the current of the actuator, the torque of the motor or other parameter adjustment modes, so as to ensure the stability of the wafer carrying process by the mechanical arm;
Referring to fig. 2, in a preferred implementation, the method for acquiring torque test data includes:
step a1: placing the test executor in a current change test environment, testing the torque value of the test executor, and placing the standard executor in a standard current test environment;
Step a2: in the test process, when the current of the test actuator is set to be p ampere, a first torque value of the actuator per newton meter is measured, and p is an integer larger than zero;
Step a3: under a standard current testing environment set by a standard actuator, acquiring a first standard torque value of each Newton meter of the standard actuator;
Step a4: taking the difference value between the first torque value and the first standard torque value as a first torque difference, comparing the first torque difference with a preset first torque error interval, if the first torque difference does not belong to the preset first torque error interval, enabling p=p+1, and returning to the step a2; if the first torque difference belongs to a preset first torque error interval, taking the first torque difference as a torque correction value, and binding and correlating a p-th ampere with the torque correction value to obtain the relation between the torque correction value and the p-th ampere current of the actuator;
Step a5: repeating the steps a2 to a4 until P is equal to the set current and the cycle is ended, obtaining the relation between the torque correction value and each ampere current of the actuator, wherein P is an integer larger than zero;
referring to fig. 3, in another embodiment, the method for acquiring torque test data further includes:
step b1: placing the test executor in a motor torque variation environment, testing the torque of the executor, and placing the standard executor in a set standard motor torque test environment;
step b2: under a test environment, when the motor torque of the test actuator is set to be w Newton-Mi Zhuaiju, a second torque value of the actuator per Newton-meter is measured, and w is an integer larger than zero;
Step b3: under a set standard motor torque test environment, obtaining a second standard torque value of each Newton meter of the standard actuator;
Step b4: taking the difference value between the second torque value and the second standard torque value as a second torque difference, comparing the second torque difference with a preset second torque error interval, if the second torque difference does not belong to the preset second torque error interval, letting w=w+1, and returning to the step b2; if the second torque difference belongs to a preset second torque error section, taking the second torque difference as a torque correction value, and binding and associating the w < Newton >. Mi Niuju with the torque correction value of the motor to obtain the relation between the torque correction value of the actuator and the w < Newton >. Meter torque of the motor;
Step b5: repeating the steps b 2-b 4 until W equals to the set torque W, ending the cycle, and obtaining the relation between the torque correction value of the actuator and each Newton meter torque of the motor, wherein W is an integer larger than zero;
in implementation, the method for generating the pre-constructed correction model is as follows:
Dividing the torque test data into a torque correction training set and a torque correction test set; constructing a machine learning model, taking the current of an actuator and the torque of a motor in a torque correction training set as input data of the machine learning model, taking the torque correction value in the torque correction training set as output data of the machine learning model, and training the machine learning model to obtain an initial correction model; performing model verification on the initial correction model by using the torque correction test set, and outputting the initial correction model meeting the preset prediction error as a pre-constructed correction model;
It should be noted that: the machine learning model can be one of a cyclic neural network, a convolutional neural network or a long-short-term memory network;
it should be appreciated that: and determining actual load data of the load sensor based on the torque value after parameter correction, wherein the actual load data is calculated according to the formula as follows:
；
wherein, Weight factor representing torque correction value,/>The weight factor representing the arm length of the mechanical arm, C is a correction constant of actual load data;
It should be noted that, the arm length of the mechanical arm is pre-stored in a system database;
S4, judging whether abnormal carrying exists, inputting the first target load data and the actual load data into a pre-constructed load correction model according to the abnormal carrying to obtain second target load data, and carrying the wafer by the mechanical arm according to the second target load data;
In an implementation, a method of determining whether there is abnormal handling includes:
Calculating a load difference value between the first target load data and the actual load data;
Comparing the load difference value with a load difference value threshold value;
If the load difference value is greater than or equal to the load difference value threshold value, judging that abnormal transportation exists;
If the load difference is smaller than the load difference threshold, judging that abnormal transportation does not exist;
Specifically, the second target load data is generated based on load test data training; the load test data acquisition method comprises the following steps:
c1: obtaining the ith abnormal handling, wherein i is an integer greater than zero;
c2: acquiring first target load data of load sensor according to ith abnormal handling And according to the first target load data/>Actual load data/>, of load sensor under control；
C3: acquiring first target load dataAnd actual load data/>Is a load error of (a);
c4: judging whether the load error is equal to zero, if the load error is not equal to zero, further judging whether the load error is greater than zero or less than zero, if the load error is less than zero, then performing data processing on the first target load Increment, let/>=/>+G, and returning to step c2; if greater than zero, for the first target load data/>Decrementing, let/>=/>-G and returning to step c2; if the load error is equal to zero, the first target load data/>, after load adjustment, is obtainedAs second target load data/>And/>, first target load dataAnd actual load data/>And second target load data/>Correlating to obtain a group of first target load data/>And actual load data/>And second target load data/>G is an integer greater than zero; /(I)
C5: repeating the steps c 2-c 4 until i=i, ending the cycle to obtain H groups of first target load dataAnd actual load data/>And second target load data/>To H groups of first target load data/>And actual load data/>And second target load data/>The relation of (3) is used as load test data, wherein I is the total number of abnormal handling, I, H is an integer greater than zero;
it should be noted that, the logic for generating the pre-constructed load correction model is as follows:
Dividing load test data into a load training set and a load test set;
Constructing a network learning model, taking first target load data and actual load data in a load training set as input data of the network learning model, taking second target load data in the load training set as output data of the network learning model, and training the network learning model to obtain an initial load correction model;
performing model verification on the initial load correction model by using a load test set, and outputting the initial load correction model meeting the preset prediction error as a pre-constructed load correction model;
it should be noted that: the network learning model is one of a convolutional neural network, a cyclic neural network or a long-short-term memory network.
According to the invention, load monitoring data of the mechanical arm in the process of carrying the wafer at the moment T is collected, a historical load evaluation data set is obtained, and a load evaluation coefficient of a load sensor at the moment T+a in the future is obtained according to a pre-built load learning model, wherein the load monitoring data comprises actual load data and first target load data; determining whether the load sensor at the moment T+a is in an abnormal load state according to the predicted load evaluation coefficient, if not, continuing to convey the wafer by the mechanical arm under the first target load data, and enabling T=T+a+b, and if so, acquiring load characteristic data of the load sensor at the moment T+a; when the load sensor is in an abnormal load state, recording a torque value of the actuator, correcting the torque value of the actuator according to load characteristic data and a pre-constructed correction model, and determining actual load data of the mechanical arm based on the corrected torque value of the actuator; the load characteristic data comprises current of an actuator and torque of a motor; calculating a load difference value between the first target load data and the actual load data, judging whether abnormal handling exists, inputting the first target load data and the actual load data into a pre-constructed load correction model according to the abnormal handling to obtain second target load data, and carrying out wafer handling by the mechanical arm according to the second target load data; according to the invention, the abnormal load judging mechanism is loaded in the mechanical arm, the analysis of the abnormal load state is carried out when the mechanical arm conveys the wafer, the torque of the actuator is corrected, the load data is adaptively adjusted based on the torque correction value of the actuator, the mechanical arm is prevented from being abnormal in the wafer conveying process, and the safety of the mechanical arm for conveying the wafer is improved.
The formulas related in the above are all formulas with dimensions removed and numerical values calculated, and are a formula which is obtained by acquiring a large amount of data and performing software simulation to obtain the closest actual situation, and weight factors in the formulas and various preset thresholds in the analysis process are set by a person skilled in the art according to the actual situation or are obtained by simulating a large amount of data; the size of the weight factor is a specific numerical value obtained by quantizing each parameter, so that the subsequent comparison is convenient, and the size of the weight factor depends on the number of sample data and the corresponding processing coefficient is preliminarily set for each group of sample data by a person skilled in the art; as long as the proportional relation between the parameter and the quantized value is not affected.
Example 2
Referring to fig. 4, the present embodiment provides a load adaptive system of a mechanical arm based on a vacuum environment, including:
The load evaluation module is used for collecting load monitoring data of the mechanical arm in the process of carrying the wafer at the moment T, acquiring a historical load evaluation data set, and acquiring a load evaluation coefficient of a load sensor at the moment T+a in the future according to a pre-constructed load learning model, wherein the load monitoring data comprises actual load data and first target load data; a is an integer greater than zero;
The load judging module is used for determining whether the load sensor at the moment T+a is in an abnormal load state according to the predicted load evaluation coefficient, if not, the mechanical arm continues to carry the wafer under the first target load data, and causes T=T+a+b, and if so, the load characteristic data of the load sensor at the moment T+a is obtained; b is an integer greater than zero;
the torque correction module is used for recording the torque value of the actuator when the load sensor is in an abnormal load state, correcting the torque value of the actuator according to the load characteristic data and a pre-constructed correction model, and determining the actual load data of the mechanical arm based on the corrected torque value of the actuator; the load characteristic data comprises current of an actuator and torque of a motor;
the load self-adaptation module is used for calculating a load difference value between the first target load data and the actual load data, judging whether abnormal handling exists, inputting the first target load data and the actual load data into a pre-constructed load correction model according to the abnormal handling, obtaining second target load data, and carrying out wafer handling by the mechanical arm according to the second target load data.
Example 3
Referring to fig. 5, the present embodiment provides an electronic device, including: a processor and a memory, wherein the memory stores a computer program for the processor to call;
the processor executes the vacuum environment based mechanical arm load adaptation method of embodiment 1 by calling a computer program stored in the memory.
Example 4
Referring to fig. 6, the present embodiment provides a computer readable storage medium storing instructions that when executed on a computer cause the computer to perform the vacuum environment based robot load adaptation method of embodiment 1.
The formulas related in the above are all formulas with dimensions removed and numerical values calculated, and are a formula which is obtained by acquiring a large amount of data and performing software simulation to obtain the closest actual situation, and weight factors in the formulas and various preset thresholds in the analysis process are set by a person skilled in the art according to the actual situation or are obtained by simulating a large amount of data; the size of the weight factor is a specific numerical value obtained by quantizing each parameter, so that the subsequent comparison is convenient, and the size of the weight factor depends on the number of sample data and the corresponding processing coefficient is preliminarily set for each group of sample data by a person skilled in the art; as long as the proportional relation between the parameter and the quantized value is not affected.
The above embodiments may be implemented in whole or in part by software, hardware, firmware, or any other combination. When implemented in software, the above-described embodiments may be implemented in whole or in part in the form of a computer program product. The computer program product comprises one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, the processes or functions described in accordance with embodiments of the present application are produced in whole or in part. The computer may be a general purpose computer, a special purpose computer, a network of computers, or other programmable devices. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, from one website site, computer, server, or data center over a wired network. The computer readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. that contains one or more sets of available media. The usable medium may be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. The semiconductor medium may be a solid state disk.
Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.
It will be clear to those skilled in the art that, for convenience and brevity of description, specific working procedures of the above-described systems, apparatuses and units may refer to corresponding procedures in the foregoing method embodiments, and are not repeated herein.
In the several embodiments provided in the present application, it should be understood that the disclosed system, apparatus and method may be implemented in other manners. For example, the apparatus embodiments described above are merely illustrative, e.g., the division of the units is merely one, and there may be additional divisions in actual implementation, e.g., multiple units or components may be combined or integrated into another system, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be an indirect coupling or communication connection via some interfaces, devices or units, which may be in electrical, mechanical or other forms.
The units described as separate units may or may not be physically separate, and units shown as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional unit in the embodiments of the present application may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit.
The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a read-only memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, or other various media capable of storing program codes.
The foregoing is merely illustrative of the present application, and the present application is not limited thereto, and any person skilled in the art will readily recognize that variations or substitutions are within the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.
Finally: the foregoing description of the preferred embodiments of the invention is not intended to limit the invention to the precise form disclosed, and any such modifications, equivalents, and alternatives falling within the spirit and principles of the invention are intended to be included within the scope of the invention.
Claims (13)
1. The vacuum environment-based mechanical arm load self-adaption method is characterized by comprising the following steps of:
collecting load monitoring data of the mechanical arm in the process of carrying the wafer at the moment T, acquiring a historical load evaluation data set, and acquiring a load evaluation coefficient of a load sensor at the moment T+a in the future according to a pre-constructed load learning model, wherein the load monitoring data comprises actual load data and first target load data; a is an integer greater than zero;
Determining whether the load sensor at the moment T+a is in an abnormal load state according to the predicted load evaluation coefficient, if not, continuing to convey the wafer by the mechanical arm under the first target load data, and enabling T=T+a+b, and if so, acquiring load characteristic data of the load sensor at the moment T+a; b is an integer greater than zero;
When the load sensor is in an abnormal load state, recording a torque value of the actuator, correcting the torque value of the actuator according to load characteristic data and a pre-constructed correction model, and determining actual load data of the mechanical arm based on the corrected torque value of the actuator; the load characteristic data comprises current of an actuator and torque of a motor;
And calculating a load difference value between the first target load data and the actual load data, judging whether abnormal handling exists, inputting the first target load data and the actual load data into a pre-constructed load correction model according to the abnormal handling to obtain second target load data, and carrying out wafer handling by the mechanical arm according to the second target load data.
2. The vacuum environment-based mechanical arm load self-adaptation method according to claim 1, wherein obtaining a load evaluation coefficient of a load sensor at a future time t+a according to a pre-constructed load learning model comprises:
Acquiring voltage difference data of a load sensor, temperature difference data of the load sensor and vibration difference data of the load sensor from the moment T-b to the moment T;
The method comprises the steps of inputting voltage difference data of a load sensor, temperature difference data of the load sensor and vibration difference data of the load sensor from time T-b to time T into a pre-built load learning model to obtain a load evaluation coefficient of the load sensor at time T+a;
Acquiring a historical load evaluation data set, wherein the historical load evaluation data set comprises voltage difference data of a load sensor, temperature difference data of the load sensor, vibration difference data of the load sensor and corresponding load evaluation coefficients in n groups of time intervals, and n is an integer larger than zero;
Dividing a historical load evaluation data set for training a load learning model into a data training set and a data testing set, constructing a first regression model, taking voltage difference data of a load sensor, temperature difference data of the load sensor and vibration difference data of the load sensor in each group of preset time intervals as inputs of the first regression model, taking a load evaluation coefficient as output of the first regression model, and training the first regression model to obtain an initial load learning model; and evaluating the model effect of the initial load learning model by using a mean square error algorithm, and screening the corresponding initial load learning model with the evaluation value larger than or equal to the preset evaluation value as a load learning model.
3. The vacuum environment-based mechanical arm load self-adaption method according to claim 2, wherein the load evaluation coefficient obtaining method is as follows:
based on the mechanical arm in a constant-speed rotation stage, acquiring a load value of a load sensor in a preset time interval; obtaining a load standard value of the load sensor in a preset time interval; the load standard value is an average load value in a plurality of preset time intervals;
extracting a load value of the load sensor in a preset time interval at each moment, and carrying out formulated calculation on the load value and a load standard value to obtain a load evaluation coefficient of the load sensor; the calculation formula of the load evaluation coefficient is as follows:
；
Wherein: representing load assessment factor,/> Representing a load value at the e-th time; /(I)And (5) representing a load standard value, wherein E is the total time in a preset time interval.
4. The vacuum environment-based robot arm load adaptation method according to claim 3, wherein the method of determining whether the load sensor at the time t+a is in an abnormal load state comprises:
comparing the load evaluation coefficient with a preset coefficient gradient threshold value; the preset coefficient gradient threshold value comprises And，/>＞/>；
If it is＞/>Judging that the load sensor at the time of T+a is in an abnormal load state;
If it is Judging that the load sensor at the time of T+a is in a normal load state;
If it is And judging that the load sensor at the time T+a is in an abnormal load state.
5. The vacuum environment-based mechanical arm load self-adaption method according to claim 4, wherein the method for correcting the torque value of the actuator according to the load characteristic data and the pre-constructed correction model comprises the following steps:
inputting the load characteristic data into a pre-constructed correction model to obtain a torque correction value;
Accumulating the torque correction value and the torque value to obtain a torque value after parameter correction;
The pre-constructed correction model is generated according to torque test data in a training way; the torque test data includes at least a relationship of a torque correction value to each ampere current of the actuator and a relationship of a torque correction value to each newton meter torque of the motor.
6. The vacuum environment-based robot arm load adaptation method according to claim 5, wherein the method of torque test data acquisition comprises:
step a1: placing the test executor in a current change test environment, testing the torque value of the test executor, and placing the standard executor in a standard current test environment;
Step a2: in the test process, when the current of the test actuator is set to be p ampere, a first torque value of the actuator per newton meter is measured, and p is an integer larger than zero;
Step a3: under a standard current testing environment set by a standard actuator, acquiring a first standard torque value of each Newton meter of the standard actuator;
Step a4: taking the difference value between the first torque value and the first standard torque value as a first torque difference, comparing the first torque difference with a preset first torque error interval, if the first torque difference does not belong to the preset first torque error interval, enabling p=p+1, and returning to the step a2; if the first torque difference belongs to a preset first torque error interval, taking the first torque difference as a torque correction value, and binding and correlating a p-th ampere with the torque correction value to obtain the relation between the torque correction value and the p-th ampere current of the actuator;
Step a5: repeating the steps a2 to a4 until P is equal to the set current and the cycle is ended, obtaining the relation between the torque correction value and each ampere current of the actuator, wherein P is an integer larger than zero;
The method for acquiring the torque test data further comprises the following steps:
step b1: placing the test executor in a motor torque variation environment, testing the torque of the executor, and placing the standard executor in a set standard motor torque test environment;
step b2: under a test environment, when the motor torque of the test actuator is set to be w Newton-Mi Zhuaiju, a second torque value of the actuator per Newton-meter is measured, and w is an integer larger than zero;
Step b3: under a set standard motor torque test environment, obtaining a second standard torque value of each Newton meter of the standard actuator;
Step b4: taking the difference value between the second torque value and the second standard torque value as a second torque difference, comparing the second torque difference with a preset second torque error interval, if the second torque difference does not belong to the preset second torque error interval, letting w=w+1, and returning to the step b2; if the second torque difference belongs to a preset second torque error section, taking the second torque difference as a torque correction value, and binding and associating the w < Newton >. Mi Niuju with the torque correction value of the motor to obtain the relation between the torque correction value of the actuator and the w < Newton >. Meter torque of the motor;
Step b5: repeating the steps b 2-b 4 until W equals to the set torque W, ending the cycle, and obtaining the relation between the torque correction value of the actuator and each Newton meter torque of the motor, wherein W is an integer larger than zero.
7. The vacuum environment-based mechanical arm load self-adaption method according to claim 6, wherein the method for generating the pre-built correction model is as follows:
Dividing the torque test data into a torque correction training set and a torque correction test set; constructing a machine learning model, taking the current of an actuator and the torque of a motor in a torque correction training set as input data of the machine learning model, taking the torque correction value in the torque correction training set as output data of the machine learning model, and training the machine learning model to obtain an initial correction model; performing model verification on the initial correction model by using the torque correction test set, and outputting the initial correction model meeting the preset prediction error as a pre-constructed correction model; the machine learning model is one of a cyclic neural network, a convolutional neural network or a long-short-term memory network;
and determining actual load data of the load sensor based on the torque value after parameter correction, wherein the calculation formula of the actual load data is as follows:
；
wherein, Weight factor representing torque correction value,/>The weight factor indicating the arm length of the robot arm, and C is a correction constant of actual load data.
8. The vacuum environment-based robot load adaptation method according to claim 7, wherein the method of determining whether abnormal handling exists comprises:
Calculating a load difference value between the first target load data and the actual load data;
Comparing the load difference value with a load difference value threshold value;
If the load difference value is greater than or equal to the load difference value threshold value, judging that abnormal transportation exists;
If the load difference is smaller than the load difference threshold, judging that abnormal conveying does not exist.
9. The vacuum environment-based robotic arm load adaptation method of claim 8, wherein the second target load data is generated based on load test data training; the load test data acquisition method comprises the following steps:
Step c1: obtaining the ith abnormal handling, wherein i is an integer greater than zero;
step c2: acquiring first target load data of load sensor according to ith abnormal handling And according to the first target load data/>Actual load data/>, of load sensor under control；
Step c3: acquiring first target load dataAnd actual load data/>Is a load error of (a);
Step c4: judging whether the load error is equal to zero, if the load error is not equal to zero, further judging whether the load error is greater than zero or less than zero, if the load error is less than zero, then performing data processing on the first target load Increment, let/>=/>+G, and returning to step c2; if greater than zero, for the first target load data/>Decrementing, let/>=/>-G and returning to step c2; if the load error is equal to zero, the first target load data/>, after load adjustment, is obtainedAs second target load data/>And/>, first target load dataAnd actual load data/>And second target load data/>Correlating to obtain a group of first target load data/>And actual load data/>And second target load data/>G is an integer greater than zero;
Step c5: repeating the steps c 2-c 4 until i=i, ending the cycle to obtain H groups of first target load data And actual load data/>And second target load data/>To H groups of first target load data/>And actual load data/>And second target load data/>As load test data, I is the total number of abnormal handling, I, H is an integer greater than zero.
10. The vacuum environment-based mechanical arm load adaptation method according to claim 9, wherein the generating logic of the pre-built load correction model is as follows:
Dividing load test data into a load training set and a load test set;
Constructing a network learning model, taking first target load data and actual load data in a load training set as input data of the network learning model, taking second target load data in the load training set as output data of the network learning model, and training the network learning model to obtain an initial load correction model;
performing model verification on the initial load correction model by using a load test set, and outputting the initial load correction model meeting the preset prediction error as a pre-constructed load correction model;
The network learning model is one of a convolutional neural network, a cyclic neural network or a long-short-term memory network.
11. A vacuum environment based robot load adaptation system for implementing the vacuum environment based robot load adaptation method of any one of claims 1-10, comprising:
The load evaluation module is used for collecting load monitoring data of the mechanical arm in the process of carrying the wafer at the moment T, acquiring a historical load evaluation data set, and acquiring a load evaluation coefficient of a load sensor at the moment T+a in the future according to a pre-constructed load learning model, wherein the load monitoring data comprises actual load data and first target load data; a is an integer greater than zero;
The load judging module is used for determining whether the load sensor at the moment T+a is in an abnormal load state according to the predicted load evaluation coefficient, if not, the mechanical arm continues to carry the wafer under the first target load data, and causes T=T+a+b, and if so, the load characteristic data of the load sensor at the moment T+a is obtained; b is an integer greater than zero;
the torque correction module is used for recording the torque value of the actuator when the load sensor is in an abnormal load state, correcting the torque value of the actuator according to the load characteristic data and a pre-constructed correction model, and determining the actual load data of the mechanical arm based on the corrected torque value of the actuator; the load characteristic data comprises current of an actuator and torque of a motor;
the load self-adaptation module is used for calculating a load difference value between the first target load data and the actual load data, judging whether abnormal handling exists, inputting the first target load data and the actual load data into a pre-constructed load correction model according to the abnormal handling, obtaining second target load data, and carrying out wafer handling by the mechanical arm according to the second target load data.
12. An electronic device, comprising: a processor and a memory, wherein the memory stores a computer program for the processor to call;
The processor performs the vacuum environment based robot arm load adaptation method of any one of claims 1-10 by invoking a computer program stored in the memory.
13. A computer readable storage medium, characterized in that instructions are stored which, when run on a computer, cause the computer to perform the vacuum environment based robot arm load adaptation method according to any of claims 1-10.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410235265.3A CN117808052B (en) | 2024-03-01 | 2024-03-01 | Vacuum environment-based mechanical arm load self-adaption method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410235265.3A CN117808052B (en) | 2024-03-01 | 2024-03-01 | Vacuum environment-based mechanical arm load self-adaption method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117808052A CN117808052A (en) | 2024-04-02 |
| CN117808052B true CN117808052B (en) | 2024-04-26 |
Family
ID=90420382
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410235265.3A Active CN117808052B (en) | 2024-03-01 | 2024-03-01 | Vacuum environment-based mechanical arm load self-adaption method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117808052B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116442250A (en) * | 2023-06-20 | 2023-07-18 | 东莞市嘉翼智能装备有限公司 | Self-adaptive intelligent control method, system and storage medium for linear motor |
| CN116810825A (en) * | 2023-08-30 | 2023-09-29 | 泓浒(苏州)半导体科技有限公司 | Wafer conveying mechanical arm abnormality monitoring method and system in vacuum environment |
| CN116853011A (en) * | 2023-07-04 | 2023-10-10 | 广汽本田汽车有限公司 | Energy recovery method, system and device for self-adaptive vehicle weight and storage medium |
| CN117407679A (en) * | 2023-12-15 | 2024-01-16 | 南京中鑫智电科技有限公司 | Data acquisition method and system of intelligent end screen sensor |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20230055079A1 (en) * | 2021-07-30 | 2023-02-23 | Samsung Electronics Co., Ltd. | Method of load forecasting via attentive knowledge transfer, and an apparatus for the same |
-
2024
- 2024-03-01 CN CN202410235265.3A patent/CN117808052B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116442250A (en) * | 2023-06-20 | 2023-07-18 | 东莞市嘉翼智能装备有限公司 | Self-adaptive intelligent control method, system and storage medium for linear motor |
| CN116853011A (en) * | 2023-07-04 | 2023-10-10 | 广汽本田汽车有限公司 | Energy recovery method, system and device for self-adaptive vehicle weight and storage medium |
| CN116810825A (en) * | 2023-08-30 | 2023-09-29 | 泓浒(苏州)半导体科技有限公司 | Wafer conveying mechanical arm abnormality monitoring method and system in vacuum environment |
| CN117407679A (en) * | 2023-12-15 | 2024-01-16 | 南京中鑫智电科技有限公司 | Data acquisition method and system of intelligent end screen sensor |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117808052A (en) | 2024-04-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US5511004A (en) | Diagnostic method for an evolutionary process | |
| EP2342603B1 (en) | Method and apparatus for creating state estimation models in machine condition monitoring | |
| CN117196353B (en) | Environmental pollution assessment and monitoring method and system based on big data | |
| CN117270514B (en) | Production process whole-flow fault detection method based on industrial Internet of things | |
| US7949497B2 (en) | Machine condition monitoring using discontinuity detection | |
| CN116678489B (en) | Quality control method, system, equipment and medium for force sensor | |
| CN116994999B (en) | Mechanical arm suction adjusting method and system for ultra-clean environment | |
| CN113424119A (en) | Work efficiency evaluation method, work efficiency evaluation device, and program | |
| CN117330963B (en) | Energy storage power station fault detection method, system and equipment | |
| CN117808052B (en) | Vacuum environment-based mechanical arm load self-adaption method and system | |
| CN112797967B (en) | Random drift error compensation method of MEMS gyroscope based on graph optimization | |
| CN112380073B (en) | Fault position detection method and device and readable storage medium | |
| CN109214447B (en) | Disk life prediction method and device | |
| CN112067289A (en) | Motor shaft and transmission shaft abnormal vibration early warning algorithm based on neural network | |
| CN116771611A (en) | Wind turbine generator power monitoring method and device | |
| US20220413480A1 (en) | Time series data processing method | |
| CN113588059B (en) | Dynamic weighing equipment parameter configuration method and system | |
| JP2004078716A (en) | Quality management system and its method | |
| CN117725541B (en) | Intelligent monitoring and fault diagnosis system for running state of annealing furnace | |
| CN112978532B (en) | Elevator state detection method and device and storage medium | |
| CN117077993B (en) | Workshop material automatic weighing data management system and method based on artificial intelligence | |
| CN115219883A (en) | Automatic detection system and detection method for electric automation equipment | |
| CN115543729A (en) | Data acquisition method and system | |
| CN118034136A (en) | Digital twin method and system applied to industrial research and development design | |
| CN114199359A (en) | Offset load error adjusting method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |