An automatic leveling method for 3D printing platform based on ANFIS fuzzy neural reasoning

Abstract

The present invention relates to the technical field of automatic leveling of 3D printing platforms, and discloses an automatic leveling method of 3D printing platforms based on ANFIS fuzzy neural reasoning, the method comprising: controlling a nozzle to make random contact on a 3D printing platform, obtaining pressure values ​​and distance values ​​of multiple corner points; training an ANFIS model according to the pressure values ​​and distance values ​​of multiple corner points obtained; controlling the nozzle to contact each corner point in turn on the 3D printing platform, obtaining the pressure value and distance value of each corner point, and obtaining a predicted Z-axis compensation value using the trained ANFIS model; controlling the nozzle to cross-contact each corner point on the 3D printing platform, and making the pressure value and distance value of each point contacted along the leveling path zero according to the predicted Z-axis compensation value. The present invention can improve the accuracy of traditional automatic leveling and leveling using a contact sensor.

Description

Technical Field

[0001] The present invention relates to the technical field of automatic leveling of a 3D printing platform, and in particular to an automatic leveling method of a 3D printing platform based on ANFIS fuzzy neural reasoning. Background Art

[0002] ANFIS (Adaptive-Network-Based Fuzzy Inference System) - Adaptive fuzzy neural network, first proposed by Jyh-Shing Roger Jang in 1993. Fuzzy if-then rules are used to establish a fuzzy inference system, which can construct input-output mapping based on human knowledge (in the form of fuzzy if-then rules) and specified input-output data pairs.

[0003] ANFIS is a fuzzy inference system based on the Takagi-Sugeno model. It uses neural networks to implement the three basic processes of fuzzification, fuzzy inference and clarification of fuzzy control. It uses the learning mechanism of neural networks to automatically extract rules from input and output sample data to form an adaptive neural fuzzy controller. It adjusts the fuzzy inference control rules through offline training and online learning algorithms, making the system itself develop in the direction of self-adaptation, self-organization and self-learning.

[0004] In the 3D printing process, automatic leveling is an important step, which ensures the appropriate distance between the print bed surface and the print nozzle, thereby ensuring the stability and consistency of the printing quality. Automatic leveling refers to the combination of software and hardware, which enables the printer to automatically detect the height difference of the print bed surface and achieve balance by adjusting the position of the print nozzle. This process is generally performed before each print to ensure the flatness of the print bed surface. If the print bed surface is not flat, the print nozzle may collide with the bed surface during printing, resulting in reduced print quality or even failure to complete printing. Summary of the invention

[0005] In view of the above-mentioned deficiencies in the prior art, the present invention provides a 3D printing platform automatic leveling method based on ANFIS fuzzy neural reasoning.

[0006] In order to achieve the above-mentioned object of the invention, the technical solution adopted by the present invention is:

[0007] A 3D printing platform automatic leveling method based on ANFIS fuzzy neural reasoning includes the following steps:

[0008] Control the nozzle to make random contact on the 3D printing platform to obtain the pressure and distance values ​​of multiple corner points;

[0009] The ANFIS model is trained based on the pressure values ​​and distance values ​​of multiple corner points;

[0010] On the 3D printing platform, the nozzle is controlled to contact each corner point in turn, the pressure value and distance value of each corner point are obtained, and the predicted Z-axis compensation value is obtained using the trained ANFIS model;

[0011] On the 3D printing platform, the nozzle is controlled to cross-contact various corner points, and the pressure value and distance value of each point contacted according to the leveling path are made zero according to the predicted Z-axis compensation value.

[0012] Preferably, the ANFIS model includes a fuzzification layer, a rule layer, a normalization layer, a clarity layer and a summation layer;

[0013] The fuzzification layer is used to convert the pressure value and the distance value into the membership degree of each fuzzy subset;

[0014] The rule layer is used to calculate the applicability of each fuzzy rule according to the membership degree of each fuzzy subset;

[0015] The normalization layer is used to calculate the normalized applicability of each fuzzy rule among all fuzzy rules according to the applicability of each fuzzy rule;

[0016] The clarification layer is used to calculate the output value of each fuzzy rule according to the normalized applicability of each fuzzy rule among all fuzzy rules;

[0017] The summation layer is used to perform accumulation processing on all fuzzy rule output values ​​to obtain the predicted Z-axis compensation value.

[0018] Preferably, the fuzzification layer converts the pressure value and the distance value into the membership degree of each fuzzy subset, specifically:

[0019] ;

[0020] in, Represents fuzzy subsets The membership degree of Represents fuzzy subsets The membership degree of Represents fuzzy subsets The membership function of Represents fuzzy subsets The membership function of Indicates the pressure value. Represents the distance value, i, j both represent the fuzzy subset number, and n represents the corner point number.

[0021] Preferably, the membership function adopts a triangular membership function, specifically:

[0022] { μ A i ( F n ) = trimf ( F n ,[ a i , b i , c i ]) μ B j ( d n ) = trimf ( d n ,[ a j , b j , c j ]) ;

[0023] in, represents the triangle membership function, Both represent antecedent parameters.

[0024] Preferably, the rule layer calculates the applicability of each fuzzy rule according to the membership of each fuzzy subset, specifically:

[0025] ;

[0026] in, represents the output value of the rule layer, represents the applicability of the kth fuzzy rule, Represents fuzzy subsets The membership function of Represents fuzzy subsets The membership function of Indicates the pressure value. Represents the distance value, i, j both represent the fuzzy subset number, and n represents the corner point number.

[0027] Preferably, the normalization layer calculates the normalized applicability of each fuzzy rule among all fuzzy rules according to the applicability of each fuzzy rule, specifically:

[0028] ;

[0029] in, represents the output value of the normalization layer, represents the applicability of the kth fuzzy rule, represents the applicability of the kth fuzzy rule, Indicates The applicability of a fuzzy subset.

[0030] Preferably, the clarification layer calculates the output value of each fuzzy rule according to the normalized applicability of each fuzzy rule among all fuzzy rules, specifically:

[0031] ;

[0032] in, represents the output value of the sharpening layer, represents the applicability of the kth fuzzy rule, represents the output value of the kth fuzzy rule, represents the applicability of the kth fuzzy rule, , , represents the consequent parameter of the fuzzy rule, Indicates the pressure value. Represents the distance value, and n represents the corner point number.

[0033] Preferably, the summation layer performs accumulation processing on all fuzzy rule output values ​​to obtain a predicted Z-axis compensation value, specifically:

[0034] ;

[0035] in, represents the output value of the summation layer, represents the applicability of the kth fuzzy rule, represents the output value of the kth fuzzy rule, Indicates the applicability of the kth fuzzy rule.

[0036] Preferably, when training the ANFIS model, a least squares estimation-gradient descent hybrid learning algorithm is used to update parameters.

[0037] Preferably, the least squares estimation-gradient descent hybrid learning algorithm is used to update the parameters as follows:

[0038] The input data is operated along the ANFIS model to the clarification layer, the antecedent parameters are fixed, and the consequent parameters are trained using the least square method; the antecedent parameters are the training parameters of the membership function in the fuzzification layer, and the consequent parameters are the output values ​​of the fuzzy rules in the clarification layer;

[0039] The input data is forwarded to the summation layer, and the antecedent parameters are trained using the gradient descent algorithm, and the loop is iterated to the global optimum.

[0040] The present invention has the following beneficial effects:

[0041] The present invention adopts ANFIS fuzzy neural network to predict the Z-axis error of 3D printing during the automatic leveling process, and uses the learning mechanism of the neural network to automatically extract rules from the pressure and distance sample data obtained by the input sensor to form an adaptive neural fuzzy controller. The fuzzy reasoning control rules are self-adjusted through offline training and online learning algorithms, so that the system itself develops in the direction of self-adaptation, self-organization, and self-learning, thereby adapting to the leveling error of the 3D printing platform itself caused by major factors such as thermal deformation of the hot bed and temperature drift of the proximity switch, and improving the accuracy of traditional automatic leveling and leveling using contact sensors. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a flow chart of a method for automatic leveling of a 3D printing platform based on ANFIS fuzzy neural reasoning in an embodiment of the present invention;

[0043] Figure 2A schematic diagram of the structure of the ANFIS model in an embodiment of the present invention;

[0044] Figure 3 Schematic diagram of a membership function in an embodiment of the present invention;

[0045] Figure 4 This is a diagram of the first round of leveling process in an embodiment of the present invention;

[0046] Figure 5 This is a diagram of the second round of leveling process in an embodiment of the present invention. DETAILED DESCRIPTION

[0047] The specific implementation modes of the present invention are described below so that those skilled in the art can understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific implementation modes. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations utilizing the concept of the present invention are protected.

[0048] like Figure 1 As shown, an automatic leveling method for a 3D printing platform based on ANFIS fuzzy neural reasoning provided by an embodiment of the present invention includes the following steps S1 to S4:

[0049] S1. Control the nozzle to make random contact on the 3D printing platform to obtain the pressure and distance values ​​of multiple corner points;

[0050] In an optional embodiment of the present invention, step S1 is performed by making random contact of the nozzle on the 3D printing platform, obtaining pressure value and distance value data of multiple points through a sensor, and performing normalization preprocessing on the data.

[0051] S2, training the ANFIS model according to the pressure values ​​and distance values ​​of multiple corner points obtained;

[0052] In an optional embodiment of the present invention, the ANFIS model constructed in step S2 includes a fuzzification layer, a rule layer, a normalization layer, a clarity layer and a summation layer;

[0053] The fuzzification layer is used to convert the pressure value and the distance value into the membership degree of each fuzzy subset;

[0054] The rule layer is used to calculate the applicability of each fuzzy rule according to the membership degree of each fuzzy subset;

[0055] The normalization layer is used to calculate the normalized applicability of each fuzzy rule among all fuzzy rules according to the applicability of each fuzzy rule;

[0056] The clarification layer is used to calculate the output value of each fuzzy rule according to the normalized applicability of each fuzzy rule among all fuzzy rules;

[0057] The summation layer is used to perform accumulation processing on all fuzzy rule output values ​​to obtain the predicted Z-axis compensation value.

[0058] In this embodiment, the multiple groups of pressure value and distance value data processed in step S1 are divided into a training set and a test set, wherein the training data set is used to learn the parameters and rules of the ANFIS model, and the test data set is used to evaluate the prediction performance of the model.

[0059] The ANFIS model constructed in this embodiment is as follows: Figure 2 As shown, the ANFIS pre-training model consists of a five-layer forward neural network structure, the first layer is a fuzzification layer, the second layer is a rule layer, the third layer is a normalization layer, the fourth layer is a clarity layer, and the fifth layer is a summation layer.

[0060] The input variable of the ANFIS model in this embodiment is the pressure value obtained by the pressure sensor installed on the nozzle of the 3D printer. And the distance value obtained by the laser ranging sensor ; The output variable of the model is the 3D printer Axis prediction coordinate value .

[0061] The first layer of the ANFIS model: fuzzification layer. The input variables are converted into the membership of each fuzzy subset. Node i in this layer is an adaptive node composed of node functions (membership functions). The parameters of the node function are rule antecedent parameters.

[0062] In the fuzzification layer, the pressure value obtained by the pressure sensor And the distance value obtained by the laser ranging sensor Set the fuzzy subsets of the 5 linguistic variable values ​​respectively and , the five fuzzy linguistic variable values ​​are negative big (NB), negative small (NS), zero (ZO), positive small (PS), and positive big (PB).

[0063] The fuzzification layer converts the pressure value and distance value into the membership degree of each fuzzy subset, specifically:

[0064] ;

[0065] in, Represents fuzzy subsets The membership degree of Represents fuzzy subsets The membership degree of Represents fuzzy subsets The membership function of Represents fuzzy subsets The membership function of Indicates the pressure value. Represents the distance value, i, j both represent the fuzzy subset number, and n represents the corner point number.

[0066] The membership function adopts the triangular membership function, which is:

[0067] { μ A i ( F n ) = trimf ( F n ,[ a i , b i , c i ]) μ B j ( d n ) = trimf ( d n ,[ a j , b j , c j ]) ;

[0068] in, represents the triangle membership function, such as Figure 3 As shown, , , , , , It is called the "precondition parameter" and is used to determine the shape of the triangular membership function. , Determine the "valley" of the triangle, parameters Determine the "peak" of the triangle. The triangle membership function is specifically:

[0069] ;

[0070] Among them, x is the input variable.

[0071] The second layer of the ANFIS model: the rule layer. The output of each node is the product of the input signals, and the result is the applicability of this rule. Generally speaking, the node functions of this layer can be other T-norm operators that perform fuzzy "and". Such operators include minimization, algebraic product, bounded product and strong product.

[0072] The rule layer calculates the applicability of each fuzzy rule according to the membership degree of each fuzzy subset, specifically:

[0073] ;

[0074] in, represents the output value of the rule layer, represents the applicability of the kth fuzzy rule, Represents fuzzy subsets The membership function of Represents fuzzy subsets The membership function of Indicates the pressure value. Represents the distance value, i, j both represent the fuzzy subset number, and n represents the corner point number.

[0075] In this embodiment, the fuzzy subset output by the fuzzification layer and Membership and In the input rule layer, each node represents a fuzzy rule, and there are fuzzy rules, and the applicability of each fuzzy rule is output by "multiplying" the membership degree. , For the The applicability of the fuzzy rules, .

[0076] The third layer of the ANFIS model: the normalization layer. For the ith node, the ratio of the fitness of the ith rule to the fitness of all rules is calculated. The output of this layer is called the normalized fitness.

[0077] The normalization layer calculates the normalized applicability of each fuzzy rule among all fuzzy rules according to the applicability of each fuzzy rule, specifically:

[0078] ;

[0079] in, represents the output value of the normalization layer, represents the applicability of the kth fuzzy rule, represents the applicability of the kth fuzzy rule, Indicates The applicability of a fuzzy subset.

[0080] In this embodiment, the applicability of the rule layer output In the input normalization layer, the weight of each fuzzy rule among all fuzzy rules is calculated, and each node outputs a normalized value. For the The normalized value of the fuzzy rules, =1 ,2 ,...,25.

[0081] The fourth layer of the ANFIS model: the clarity layer. Calculate the output of each rule. It is the third layer of standardized applicability.

[0082] The clarity layer calculates the output value of each fuzzy rule according to the normalized applicability of each fuzzy rule among all fuzzy rules, specifically:

[0083] ;

[0084] in, represents the output value of the sharpening layer, represents the applicability of the kth fuzzy rule, represents the output value of the kth fuzzy rule, represents the applicability of the kth fuzzy rule, , , represents the consequent parameter of the fuzzy rule, Indicates the pressure value. Represents the distance value, and n represents the corner point number.

[0085] In this embodiment, the normalized value output by the normalization layer is input into the clarification layer. In this layer, each node outputs a fuzzy rule output value, and each fuzzy rule output value is a constant. , =1,2,...,125; where the parameter It is called "afterthought parameter".

[0086] The fifth layer of the ANFIS model: the summation layer, which sums each output to get the total output.

[0087] The summation layer accumulates all the fuzzy rule output values ​​to obtain the predicted Z-axis compensation value, which is:

[0088] ;

[0089] in, represents the output value of the summation layer, represents the applicability of the kth fuzzy rule, represents the output value of the kth fuzzy rule, Indicates the applicability of the kth fuzzy rule.

[0090] In this embodiment, the output of all fuzzy rules is accumulated in the summation layer, that is, the normalized value in the normalization layer is summed with the output value of the corresponding defuzzification layer. After multiplication and summing, we get the final output of ANFIS 3D printer. Axis prediction coordinate value .

[0091] When training the ANFIS model, this embodiment first determines the fuzzy partition and fuzzy set type of the input data, and then uses a hybrid algorithm of the back propagation algorithm and the linear least squares method to train the antecedent parameters and the consequent parameters. The antecedent parameters are trained using the back propagation algorithm, and the consequent parameters are trained using the least squares method. The input data is calculated along the ANFIS network to the fourth layer, the antecedent parameters are fixed, the consequent parameters are trained using the least squares method, the signal is forwarded to the output layer, and the antecedent parameters are trained using the back propagation algorithm, and the cycle is iterated to the global optimum.

[0092] The prediction performance of the ANFIS model when training in this embodiment is measured by three indicators: root mean square error (RMSE), mean absolute error (MAE), and average percentage error (APE). If the above indicators are met, the next step is executed. If not, the ANFIS model parameters, i.e., the antecedent parameters and the consequent parameters of the ANFIS model, are adjusted. The learning algorithm (parameter updating method) proposed in the present invention is the LSE-GD hybrid learning algorithm. That is, the update parameters are performed in both the forward pass and the backward pass. In the back propagation, the backward parameters (parameters in the rule) are fixed, and the parameters are updated from the opposite direction of the gradient direction according to the partial derivative of the Loss function to the forward parameters (using the chain rule).

[0093] After determining the antecedent parameters , , , , , After that, the input / output data can be used to identify the subsequent parameters using the least squares method (LSE):

[0094] ;

[0095] in, for The matrix of for The consequent parameter vector of is the random variable difference of the actual training point. The consequent parameters can be identified by the least squares method. , let the error index function be .

[0096] According to the least squares principle, At least, there must be:

[0097] ;

[0098] in, for , thereby obtaining the optimized model post-parameters .

[0099] Fixing the post-processing parameters of the ANFIS model , using the back-propagation gradient descent (GD) algorithm to reversely learn to adjust the antecedent parameters , , , , , , considering the error indicator function , yes The current output at time, is the expected output of the ANFIS model. The adjustment algorithm can be expressed as:

[0100] ;

[0101] in , , are the antecedent parameters of the ANFIS model , , The corresponding learning rate. According to the chain rule of partial derivatives, we get: ; ; .

[0102] in , as long as each iteration, change the parameter , Ep can be made to develop in the direction of reduction, that is, the output value of the model can be closer to the true output value.

[0103] It should be understood by those skilled in the art that the ANFIS pre-trained model refers to a model that has not been compared with a test set. In order to simulate the generalization error, the data set is divided into a training set and a test set, the model is trained on the training set, and then the model is tested on the test set. Compare and evaluate the test set data, and use the error value obtained on the test set as an approximate estimate of the true generalization error. With this error value as a reference, a model with a smaller error on the test set can be screened out as the final result output. The goal of training is to minimize the error value of the model on the test set. Obtain the final ANFIS prediction model. It should be understood by those skilled in the art that the final ANFIS prediction model refers to a model that has been compared with a test set, that is, a model with a smaller error on the test set is output as the final result, has a higher accuracy, and can be applied to actual predictions.

[0104] S3, controlling the nozzle to contact each corner point in turn on the 3D printing platform, obtaining the pressure value and distance value of each corner point, and using the trained ANFIS model to obtain the predicted Z-axis compensation value;

[0105] In an optional embodiment of the present invention, step S3 performs a first round of leveling, such as Figure 4 As shown, the nozzle touches each corner point of the platform, and moves the nozzle to another corner in the order of 1, 2, 3, and 4 corners, and repeats the above operation until the leveling of the four corners is completed. The static pressure value and distance value of each corner point are obtained, and the predicted Z-axis compensation value is obtained through the trained final ANFIS prediction model.

[0106] S4. Control the nozzle to cross-contact each corner point on the 3D printing platform, and make the pressure value and distance value of each point contacted along the leveling path zero according to the predicted Z-axis compensation value.

[0107] In an optional embodiment of the present invention, step S4 performs a second round of leveling, such as Figure 5 As shown, the cross-order leveling of 1, 3, 2, and 4 is adopted to easily achieve the leveling effect. According to the Z-axis compensation value predicted by the model, the pressure value and distance value of each point contacted along the leveling path are made zero, and the automatic leveling process is finally completed.

[0108] In summary, the present invention provides a method for automatic leveling of a 3D printing platform based on ANFIS fuzzy neural reasoning. The method proposes the use of ANFIS fuzzy neural network to predict the Z-axis error of 3D printing during the automatic leveling process, and uses the learning mechanism of the neural network to automatically extract rules from the pressure and distance sample data obtained by the input sensor to form an adaptive neural fuzzy controller. The fuzzy reasoning control rules are self-adjusted through offline training and online learning algorithms, so that the system itself develops in the direction of self-adaptation, self-organization, and self-learning, thereby adapting to the leveling error caused by the main factors of the 3D printing platform itself due to thermal deformation of the hot bed, temperature drift of the proximity switch, etc., and improving the accuracy of traditional automatic leveling and leveling using contact sensors.

[0109] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0110] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0111] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0112] The present invention uses specific embodiments to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

[0113] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the present invention.

Claims

1. A 3D printing platform automatic leveling method based on ANFIS fuzzy neural reasoning, characterized in that: The following steps are involved: Control the nozzle to make random contact on the 3D printing platform to obtain the pressure and distance values ​​of multiple corner points; An ANFIS model is trained according to the pressure values ​​and distance values ​​of the plurality of corner points obtained; the ANFIS model comprises a fuzzification layer, a rule layer, a normalization layer, a clarity layer and a summation layer; The fuzzification layer is used to convert the pressure value and the distance value into the membership degree of each fuzzy subset; The rule layer is used to calculate the applicability of each fuzzy rule according to the membership degree of each fuzzy subset; The normalization layer is used to calculate the normalized applicability of each fuzzy rule among all fuzzy rules according to the applicability of each fuzzy rule; The clarification layer is used to calculate the output value of each fuzzy rule according to the normalized applicability of each fuzzy rule among all fuzzy rules; The summation layer is used to accumulate all fuzzy rule output values ​​to obtain a predicted Z-axis compensation value; On the 3D printing platform, the nozzle is controlled to contact each corner point in turn, the pressure value and distance value of each corner point are obtained, and the predicted Z-axis compensation value is obtained using the trained ANFIS model; On the 3D printing platform, the nozzle is controlled to cross-contact various corner points, and the pressure value and distance value of each point contacted according to the leveling path are made zero according to the predicted Z-axis compensation value.

2. According to claim 1, a 3D printing platform automatic leveling method based on ANFIS fuzzy neural reasoning is characterized in that: The fuzzification layer converts the pressure value and the distance value into the membership degree of each fuzzy subset, specifically: ； in, Represents fuzzy subsets The membership degree of Represents fuzzy subsets The membership degree of Represents fuzzy subsets The membership function of Represents fuzzy subsets The membership function of Indicates the pressure value. Represents the distance value, i, j both represent the fuzzy subset number, and n represents the corner point number.

3. The method for automatic leveling of a 3D printing platform based on ANFIS fuzzy neural reasoning according to claim 2 is characterized in that: The membership function adopts a triangular membership function, specifically: ； in, represents the triangle membership function, Both represent antecedent parameters.

4. The method for automatic leveling of a 3D printing platform based on ANFIS fuzzy neural reasoning according to claim 1, characterized in that: The rule layer calculates the applicability of each fuzzy rule according to the membership degree of each fuzzy subset, specifically: ； in, represents the output value of the rule layer, represents the applicability of the kth fuzzy rule, Represents fuzzy subsets The membership function of Represents fuzzy subsets The membership function of Indicates the pressure value. Represents the distance value, i, j both represent the fuzzy subset number, and n represents the corner point number.

5. The method for automatic leveling of a 3D printing platform based on ANFIS fuzzy neural reasoning according to claim 1, characterized in that: The normalization layer calculates the normalized applicability of each fuzzy rule among all fuzzy rules according to the applicability of each fuzzy rule, specifically: ； in, represents the output value of the normalization layer, represents the applicability of the kth fuzzy rule, represents the applicability of the kth fuzzy rule, Indicates The applicability of a fuzzy subset.

6. The method for automatic leveling of a 3D printing platform based on ANFIS fuzzy neural reasoning according to claim 1, characterized in that: The clarification layer calculates the output value of each fuzzy rule according to the normalized applicability of each fuzzy rule among all fuzzy rules, specifically: ； in, represents the output value of the sharpening layer, represents the applicability of the kth fuzzy rule, represents the output value of the kth fuzzy rule, represents the applicability of the kth fuzzy rule, , , represents the consequent parameter of the fuzzy rule, Indicates the pressure value. Represents the distance value, and n represents the corner point number.

7. The method for automatic leveling of a 3D printing platform based on ANFIS fuzzy neural reasoning according to claim 1, characterized in that: The summation layer accumulates all fuzzy rule output values ​​to obtain the predicted Z-axis compensation value, which is specifically: ； in, represents the output value of the summation layer, represents the applicability of the kth fuzzy rule, represents the output value of the kth fuzzy rule, Indicates the applicability of the kth fuzzy rule.

8. The method for automatic leveling of a 3D printing platform based on ANFIS fuzzy neural reasoning according to claim 1, characterized in that: When training the ANFIS model, the least squares estimation-gradient descent hybrid learning algorithm is used to update parameters.

9. The method for automatic leveling of a 3D printing platform based on ANFIS fuzzy neural reasoning according to claim 8, characterized in that: The least squares estimation-gradient descent hybrid learning algorithm is used to update parameters as follows: The input data is operated along the ANFIS model to the clarification layer, the antecedent parameters are fixed, and the consequent parameters are trained using the least square method; the antecedent parameters are the training parameters of the membership function in the fuzzification layer, and the consequent parameters are the output values ​​of the fuzzy rules in the clarification layer; The input data is forwarded to the summation layer, and the antecedent parameters are trained using the gradient descent algorithm, and the loop is iterated to the global optimum.

Images