Sand blasting machine intelligent control method and system based on process parameter self-adaption

Abstract

The present application proposes a process parameter self-adaptive sand blasting machine intelligent control method and system, the method comprising: collecting sand blasting pressure, processing time and environmental humidity, inputting an artificial intelligence surface effect estimation model to generate a current surface effect estimation value; generating a control deviation value containing directional enhancement by comparing the target surface standard value; calculating the pressure channel distribution coefficient according to the deviation value, and distributing it as pressure and time adjustment amount to drive the actuator to close-loop adjustment. The present scheme solves the problem of large environmental interference and regulation lag of the existing sand blasting surface quality, realizes high-precision adaptive control of the matt sand feeling of pure titanium appearance parts, and has the beneficial effects of no additional detection hardware, high control precision and strong robustness.

Description

Technical Field

[0001] This invention belongs to the field of industrial control, and in particular relates to an intelligent control method and system for sandblasting machines based on adaptive process parameters. Background Technology

[0002] Sandblasting is a common and fundamental process in metal surface manufacturing, especially in pure titanium and its alloy products, appliance exterior parts, and consumer electronics structural components. The sandblasting process not only cleans, activates, and corrects micro-defects, but also directly determines the final matte finish, gloss uniformity, tactile smoothness, and batch-to-batch appearance consistency. For industrial products where appearance quality is a core indicator, process parameters such as sandblasting pressure and processing time directly affect the blasting intensity and cumulative effect. Furthermore, ambient humidity further influences the medium state and the actual blasting effect, resulting in different surface outcomes under the same process settings and different production conditions.

[0003] In current production, most sandblasting equipment still relies on experience-based settings or fixed process formulas. Parameters are determined during the sample testing phase and then used in mass production. When the surface effect deviates from the requirements, operators adjust the pressure or extend the processing time based on experience. Although some technical solutions introduce parameter acquisition, effect prediction, or threshold control mechanisms, they still have significant limitations in practical applications. On the one hand, the target state of a pure titanium sandblasted surface is often difficult to obtain in a timely manner through direct on-site measurement, resulting in an unstable judgment of the current surface state in the control process and making it difficult to form a unified evaluation basis consistent with the target surface standard. On the other hand, the effects of sandblasting pressure and processing time on the surface effect are not simply linearly additive; the adjustment priorities of the two are not the same in different deviation ranges. Existing technologies generally lack a continuous control chain from the current process state to surface effect estimation, from surface effect estimation to control deviation expression, and from control deviation expression to specific corrective actions applied to sandblasting pressure and processing time. Therefore, problems such as unclear adjustment basis, unstable parameter correction direction, insufficient convergence ability near the target, and poor batch consistency are prone to occur. Especially in the scenario of controlling the matte sand texture on the surface of pure titanium, the material itself is quite sensitive to the sandblasting effect. It requires both rapid correction when the deviation is large and maintaining fine and stable adjustment when approaching the target state. This makes it difficult for the existing technical route that relies on experience-based callback or simple proportional correction to simultaneously take into account response speed, control stability and surface consistency. Summary of the Invention

[0004] This invention discloses an intelligent control method and system for sandblasting machines based on adaptive process parameters, in order to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the first aspect of the present invention provides an intelligent control method for a sandblasting machine based on adaptive process parameters, the method comprising: The current sandblasting pressure, processing time, and ambient humidity of the sandblasting machine are collected and input into the artificial intelligence surface effect estimation model pre-deployed in the control unit to generate the current surface effect estimate. Obtain the pre-stored target surface standard value, compare the current surface effect estimate with the target surface standard value, and generate a control deviation value that includes deviation direction, deviation degree, and directional enhancement information; Calculate the pressure channel allocation coefficient based on the control deviation value, and allocate the control deviation value as sandblasting pressure adjustment amount and processing time adjustment amount according to the pressure channel allocation coefficient; The sandblasting pressure adjustment and the processing time adjustment are respectively added to the current sandblasting pressure setting and the current processing time setting to generate updated sandblasting pressure setting and updated processing time setting, and the sandblasting machine actuator is driven to perform sandblasting according to the updated sandblasting pressure setting and the updated processing time setting.

[0006] Furthermore, generating the current surface effect estimate specifically includes: The collected sandblasting pressure, processing time, and ambient humidity are normalized, and the normalized values ​​are used to form an input vector to input the artificial intelligence surface effect estimation model. The artificial intelligence surface effect estimation model is a feedforward neural network containing an input layer, two hidden layers and an output layer. The single-value result output by the output layer is the current surface effect estimate.

[0007] Further, the step of comparing the current surface effect estimate with the target surface standard value to generate a control deviation value containing information on deviation direction, deviation degree, and directional enhancement specifically includes: The original surface deviation is obtained by subtracting the target surface standard value from the current surface effect estimate; The original surface deviation is introduced with a target neighborhood smoothing term and a directionality enhancement term to generate the control deviation value after nonlinear shaping. Specifically, when the original surface deviation is greater than zero, the amplitude of the control deviation increases as the deviation increases; when the original surface deviation is less than zero, no directionality enhancement is performed.

[0008] Furthermore, the calculation of the pressure channel allocation coefficient based on the control deviation value specifically includes: The control deviation value is processed using a saturated normalization function that includes directional allocation gain and allocation smoothing coefficient, so that the pressure channel allocation coefficient falls within the range of 0 to 1. Specifically, when the control deviation value is greater than zero, the pressure channel allocation coefficient is greater than the basic average proportion; when the control deviation value is less than zero, the pressure channel allocation coefficient is less than the basic average proportion.

[0009] Furthermore, the step of allocating the control deviation value as a sandblasting pressure adjustment amount and a processing time adjustment amount according to the pressure channel allocation coefficient specifically includes: The sandblasting pressure adjustment amount is obtained by multiplying the product of the pressure channel allocation coefficient and the pressure channel proportional coefficient and the control deviation value. The processing time adjustment amount is obtained by multiplying the product of the pressure channel allocation coefficient and the time channel proportional coefficient and the control deviation value. The directions of the sandblasting pressure adjustment and the processing time adjustment are opposite to the direction of the control deviation value.

[0010] Furthermore, the step of adding the sandblasting pressure adjustment amount and the processing time adjustment amount to the current sandblasting pressure setting value and the current processing time setting value, respectively, specifically involves: The sandblasting pressure adjustment amount is multiplied by the pressure channel smoothing coefficient and then added to the current sandblasting pressure setting value to generate the updated sandblasting pressure setting value. The updated processing time setting is generated by multiplying the processing time adjustment by the time channel smoothing coefficient and adding it to the current processing time setting.

[0011] Furthermore, the drive mechanism of the sandblasting machine performs sandblasting according to the updated sandblasting pressure setting and the updated processing time setting, specifically including: The updated sandblasting pressure setting is converted into a control current signal to drive the electromagnetic pressure regulating valve or proportional valve to adjust the valve core opening of the air supply branch. Write the updated processing time setting into the sandblasting timing execution module to adjust the spray gun dwell time or the conveyor speed.

[0012] Furthermore, the training method for the artificial intelligence surface effect estimation model is as follows: Pure titanium samples with consistent initial surface conditions were selected. Under the condition of keeping the spray gun structure, spray distance and sand specifications consistent, the combination of sandblasting pressure, processing time and ambient humidity was changed to form multiple working conditions. After sandblasting under each working condition, the surface of the sample is tested offline and a comprehensive surface evaluation value is generated. The neural network is trained under supervision using the multiple sets of working conditions as input samples and the comprehensive surface evaluation value as the output label.

[0013] Furthermore, obtaining the pre-stored target surface standard value specifically includes: Pure titanium standard samples that meet appearance requirements were prepared using stable sandblasting conditions; The pure titanium standard sample was subjected to offline testing, and the offline testing results were converted into single-value surface evaluation results according to the evaluation rules consistent with those used when training the artificial intelligence surface effect estimation model.

[0014] A second aspect of the invention provides an intelligent control system for a sandblasting machine based on adaptive process parameters, the system comprising: The effect estimation module is used to collect the current sandblasting pressure, processing time, and ambient humidity of the sandblasting machine; input the pre-deployed artificial intelligence surface effect estimation model, and generate the current surface effect estimate value based on the data collected by the process parameter acquisition module; The deviation generation module is used to store the target surface standard value and generate a control deviation value based on the current surface effect estimate and the target surface standard value. The parameter correction module is used to calculate the pressure channel allocation coefficient based on the control deviation value, and generate the sandblasting pressure adjustment amount and the processing time adjustment amount. The execution control module is used to add the sandblasting pressure adjustment amount and the processing time adjustment amount to the current set value to generate an updated set value, and drive the sandblasting machine to perform sandblasting processing according to the updated set value.

[0015] The beneficial technical effects of the present invention are at least as follows: To address the aforementioned issues, this invention provides an intelligent control method and system for sandblasting machines based on adaptive process parameters. By deploying an artificial intelligence surface effect estimation model, it directly predicts the surface effect using conventional process parameters such as sandblasting pressure, processing time, and ambient humidity, eliminating the need for additional expensive surface detection hardware and effectively reducing system cost and complexity. By introducing a control deviation value generation mechanism with directional enhancement terms and a pressure channel allocation coefficient, it achieves nonlinear and differentiated dynamic adjustment of sandblasting pressure and processing time, solving the problems of lag, overshoot, and insufficient adjustment accuracy of traditional PID control when facing nonlinear processes, significantly improving the consistency and stability of sandblasted surface quality. This method is particularly suitable for processing pure titanium exterior parts with stringent requirements for matte finish, roughness, and gloss, offering advantages such as high control accuracy, strong adaptability, and ease of engineering deployment. Attached Figure Description

[0016] The present invention will be further described with reference to the accompanying drawings, but the embodiments in the drawings do not constitute any limitation on the present invention. For those skilled in the art, other drawings can be obtained based on the following drawings without creative effort.

[0017] Figure 1 This is a flowchart of the intelligent control method for sandblasting machines based on adaptive process parameters according to the present invention.

[0018] Figure 2 This is a framework diagram of the intelligent control system for a sandblasting machine based on adaptive process parameters, as described in this invention. Detailed Implementation

[0019] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0020] In one or more embodiments, such as Figure 1 As shown, a method for intelligent control of a sandblasting machine based on adaptive process parameters is disclosed, the method comprising the following: S1: Collect the current sandblasting pressure, processing time, and ambient humidity of the sandblasting machine, and input them into the artificial intelligence surface effect estimation model pre-deployed in the control unit to generate the current surface effect estimate.

[0021] Specifically, during the sandblasting process, the current process status is first extracted from the sandblasting machine control link and then converted into a surface effect estimate that can be directly used for subsequent deviation judgment. In practice, the sandblasting pressure is collected by a pressure detection element on the air supply branch of the sandblasting machine. After the detection element outputs an analog electrical signal, it is sent to the analog-to-digital conversion channel of the controller, which reads it into the register according to a fixed sampling period. The processing time is started by the sandblasting machine control program when the workpiece enters the sandblasting zone and stopped when the workpiece leaves the sandblasting zone or the current processing segment ends; therefore, this value is directly generated by the internal counter of the controller. The ambient humidity is collected by a humidity detection element located near the sandblasting station, preferably installed at the air inlet of the sandblasting box or in a stable airflow area outside the sandblasting station to avoid sandblasting dust adhesion causing reading drift. These three data points are recorded as follows: , and ,in The sandblasting pressure is indicated by a pressure sensing element sampled by the controller. This indicates the processing time, generated by the controller's internal timing module. This indicates the ambient humidity, which is collected by a humidity detection element. Then, the upper and lower limit parameters pre-written into the control unit during the equipment commissioning phase are invoked. , , Linear normalization is performed to ensure that the three inputs fall within a uniform numerical range. The upper and lower limits mentioned here are derived from the process window calibration during the trial production phase of the sandblasting equipment. For example, when sandblasting pure titanium workpieces, the spray gun structure, blasting medium, and spray distance are fixed first. Multiple test blasts are performed within the allowable process range, and the minimum and maximum values ​​at which stable forming occurs are recorded. This range is then written into the control program. After normalization, the three inputs form the input vector. ,in Indicates the normalized , , The column vectors formed by arranging them sequentially represent the unified expression of the current working condition at the model input.

[0022] Surface effect estimation is achieved using a pre-trained feedforward neural network deployed in the control unit. The model structure consists of an input layer with 3 nodes, a first hidden layer with 8 nodes, a second hidden layer with 4 nodes, and an output layer with 1 node. The first hidden layer is used to extract the main coupling relationship between blasting pressure and processing time, because in the blasting of pure titanium surfaces, these two quantities jointly determine the degree of impact on the surface. The second hidden layer is used to compress and reconstruct the output of the first hidden layer, so that the modulating effect of ambient humidity on the aforementioned coupling relationship can be expressed independently. The output layer provides a single-value result, denoted as the surface effect estimate. ,in This represents the surface effect estimate output by the model, used to characterize the overall matte and sandy feel of a pure titanium surface under the current sandblasting conditions. The model employs a layer-by-layer calculation principle of "linear transformation plus nonlinear activation" in neural networks. Each layer first performs matrix multiplication and bias addition, then introduces nonlinearity through an activation function. In this embodiment, the input is limited to three data points: sandblasting pressure, processing time, and ambient humidity. The output is limited to a single-value surface effect estimate serving sandblasting control. The model calculation is expressed as follows: ; in, This represents the weight matrix from the input layer to the first hidden layer, with 8 rows and 3 columns. This represents the bias column vector of the first hidden layer, with a length of 8; This represents the weight matrix from the first hidden layer to the second hidden layer, with 4 rows and 8 columns; This represents the bias column vector of the second hidden layer, with a length of 4; This represents the weight matrix from the second hidden layer to the output layer, with 1 row and 4 columns. Indicates the bias of the output layer; This represents the ReLU activation operation, which evaluates the intermediate result of each neuron one by one. If the result is greater than 0, the original value is retained; if it is less than or equal to 0, it is set to 0. The interlayer relationship in this formula is as follows: the first layer extracts basic coupling features from the three inputs; the second layer compresses these basic coupling features into combined features suitable for controlling the decision; and finally, the output layer maps the combined features into continuous numerical values. Model parameters , , , , , The training samples were obtained through sample training before the equipment went live. The training samples were obtained by selecting pure titanium samples from the same batch or with controlled differences in initial surface conditions, keeping the spray gun structure, spray distance, and abrasive specifications consistent, and only changing the combination of blasting pressure, processing time, and ambient humidity to form multiple working conditions. After blasting each working condition, the sample surface was offline inspected, and the inspection results were compressed into a comprehensive surface evaluation value according to preset evaluation rules. This comprehensive surface evaluation value was then used as a supervisory signal to train the model parameters. Therefore, the output during the operational phase... The corresponding surface effect characterization value is obtained by learning the sample relationship between "process conditions and surface results" in the pure titanium sandblasting experiment.

[0023] In actual operation, if the normalized sandblasting pressure, processing time, and ambient humidity are 0.60, 0.50, and 0.40 at a certain moment, respectively, then the input vector... Written as Let the weight behavior of the first neuron in the first hidden layer be... With a bias of 0.1, the linear combination result of this neuron can be written as: ; in, This represents the intermediate result of the first neuron in the first hidden layer before linear activation; where 0.8, 0.5, and -0.3 are the values ​​of this neuron for the input vector. The connection weights of the three components are given, and 0.1 is the bias term for this neuron. Because... If the result is greater than 0, the neuron's output remains 0.71 after ReLU. The remaining 7 neurons are calculated in the same way, resulting in an 8-dimensional intermediate vector. The second hidden layer then performs the same "weighted summation-bias correction-ReLU processing" on this 8-dimensional intermediate vector to obtain a 4-dimensional combined feature. The output layer then processes these 4 values ​​according to... and After performing a linear combination, a single-value output is obtained. For example, a calculation result of 0.67 represents the estimated surface effect under the current sandblasting condition. The value is 0.67. The controller performs a forward calculation according to this process in each sampling cycle, thereby continuously outputting the estimated surface effect value corresponding to the current operating condition. This value will proceed directly to the next step, where it will be compared with the target surface standard to generate a control deviation value.

[0024] S2: Obtain the pre-stored target surface standard value, compare the current surface effect estimate with the target surface standard value, and generate a control deviation value that includes deviation direction, deviation degree and directional enhancement information.

[0025] Specifically, step two directly receives the surface effect estimate output from step one. Furthermore, under the same evaluation scale as the target surface standard, it is converted into a control deviation value that can be directly incorporated into subsequent parameter correction calculations. In specific implementation, the target surface standard value is pre-stored within the sandblasting machine control unit. This value comes from the pure titanium standard sample during the product prototyping stage: first, a standard sample meeting the appearance requirements was prepared using stable sandblasting conditions; then, following the same evaluation rule used in step one when training the artificial intelligence surface effect estimation model, the offline detection result of this standard sample was converted into a single-value surface evaluation result, and this result was written into the control unit as... .so, and Both correspond to the same surface effect evaluation space. The former is output online by the artificial intelligence surface effect estimation model in step one, while the latter is calibrated offline by standard samples and then stored. The two can be directly compared. For the pure titanium matte sandblasting scenario, using only the linear difference between the current value and the target value can reflect the deviation direction, but it is prone to insufficient control resolution when approaching the target. When the deviation is large, it will cause the subsequent parameter adjustment to be too drastic. Therefore, this step introduces a target neighborhood smoothing term and a directionality enhancement term on the basis of the classic error definition, and converts the surface effect estimate into a deviation signal that is more suitable for the adaptive control of the sandblasting machine process parameters.

[0026] First, based on the error definition in metrology and classical control theory, the original surface deviation is constructed from the current surface effect estimate and the target surface standard value. The original source of this definition is the error expression "difference between the current value and the reference value." This step uses this as the first-level foundational quantity to preserve the direction and degree of deviation of the sandblasting result relative to the target surface condition. Its calculation relationship is as follows: ; in, Indicates the original surface deviation; This indicates that the surface effect estimate output from step one is obtained by forward calculation of sandblasting pressure, processing time, and ambient humidity using an artificial intelligence surface effect estimation model. The target surface standard value is derived from a pure titanium standard sample, measured offline, and pre-calculated according to the evaluation rules consistent with step one before being written into the control unit. This formula gives... It already contains the basic directional information required for subsequent control: when A value greater than zero indicates that the current surface effect estimate is higher than the target surface standard; when A value less than zero indicates that the current surface effect estimate is lower than the target surface standard; when When the deviation is close to zero, it indicates that the current sandblasting state is close to the target matte finish. Based on this initial deviation, this step further introduces nonlinear shaping. The idea originates from the saturated error compression structure in control theory, which limits the magnitude of the error to maintain fine-tuning capability within the target neighborhood and stable output within a larger deviation range. Simultaneously, considering the actual process characteristics of pure titanium sandblasting surfaces, a directional enhancement term is added to the side with positive deviation to improve adjustment sensitivity under conditions of strong surface interaction. Based on this, the control deviation value... Calculate using the following formula: ; in, This represents the control deviation value, which serves as the direct input for generating adaptive correction values ​​for process parameters in the next step. The target neighborhood smoothing coefficient is determined by the pure titanium test spraying experiment during the equipment commissioning phase and written into the control unit to ensure that when... When the denominator is close to zero, it remains positive and the deviation shaping process is continuous. The directional enhancement coefficient, obtained from the calibration of "the required adjustment sensitivity under the condition of strong surface effect" in the pure titanium surface test spraying, is used to enhance the control weight in the positive deviation range. The first part of the formula... It is derived from classical linear error The derivation logic of the saturated bias compression term obtained from the derivation is: retaining the sign and direction of the error, in order to... This participates in the denominator construction, causing the output to gradually enter the controlled range as the deviation increases; in the second part... This is derived from the idea of ​​symbolic choice, when When positive, this term takes a positive value and increases as the deviation increases; when... When negative, this term is zero, thus the directionality enhancement only acts on the side where the surface effect is higher than the target value. This term is then compared with the coefficient... After multiplying and enclosing the results in parentheses, we obtain the asymmetric control deviation expression applicable to pure titanium blasting scenarios. With this construction, It has three functions at the same time: using The deviation direction is retained, the amplitude change is controlled by fractional compression terms, and the directional enhancement terms are used to express the process characteristics of pure titanium surfaces that require more sensitive correction under strong working conditions.

[0027] Applying the above calculations to specific working conditions yields the complete implementation process. Let the estimated surface effect value output in step one at a certain moment be... The target surface standard value is pre-stored in the control unit The original surface deviation is obtained from the first equation. Let's re-establish the target neighborhood smoothing coefficient determined during the debugging phase. Directional enhancement coefficient When substituting into the second equation, first calculate... The first part is The fractional terms in the second part are: Therefore, the result in parentheses is Finally obtained This result indicates that the current surface finish is below the target value. The next step generates process parameter corrections to enhance the blasting effect. If, at another time, the estimated surface finish output from step one is... Then we get from the first equation ,at this time The first part is The fractional terms in the second part are: The result in parentheses is Finally obtained The results indicate that the current surface effect is higher than the target value, and because it is on the stronger side, the amplitude of the control deviation value is larger after directional enhancement. The next step will more sensitively generate adaptive correction amounts for process parameters to reduce the sandblasting effect. These two sets of calculations show that the constructed... It retains the ability to change continuously near the target, maintains directional stability when there is a significant deviation, and gives a higher control weight to the side with stronger effect on the pure titanium sandblasting surface.

[0028] The only output of step two is the control deviation value. This result is not a simple difference, but rather an estimate of the surface effect output from step one. The control variable obtained after continuous deduction through "original deviation construction - nonlinear shaping - directionality enhancement" has thus uniformly encoded the deviation direction and degree between the current sandblasting state and the target surface state, as well as the asymmetric process sensitivity under the pure titanium sandblasting scenario, into a single variable. The next step is to directly use... As input, the adaptive correction amount of the process parameters of sandblasting pressure and processing time is calculated, so that the control deviation value formed in this step becomes a key intermediate quantity in the entire intelligent control chain of the sandblasting machine.

[0029] S3: Calculate the pressure channel allocation coefficient based on the control deviation value, and allocate the control deviation value as sandblasting pressure adjustment amount and processing time adjustment amount according to the pressure channel allocation coefficient.

[0030] Specifically, step three directly receives the control deviation value output from step two. This value is then converted into an adaptive correction value for the process parameters that the sandblasting machine control unit can directly execute. The core of this step is not... Instead of mechanically amplifying and applying the sandblasting pressure and processing time separately, this step focuses on the process characteristics of pure titanium sandblasting, mapping both the "direction of the current surface state deviating from the target" and the "adjustment stage of the deviation" to two execution channels simultaneously. This is because the response of a pure titanium surface to blasting intensity and duration is not consistent during sandblasting: when the surface effect is higher than the target surface standard, prioritizing a reduction in sandblasting pressure is more effective in quickly weakening the intensity of a single impact; when the surface effect is lower than the target surface standard but the deviation is small, appropriately extending the processing time is more beneficial for fine compensation and maintaining surface uniformity. Therefore, this step first... Generate pressure channel allocation coefficient Then, using this allocation coefficient, the same control deviation value is split into sandblasting pressure adjustment amounts. Adjustment amount with processing time .in, This indicates the correction amount for the sandblasting pressure. The control unit uses this value to correct the set value of the solenoid pressure regulating valve or proportional valve. This represents the correction amount for processing time. The control program uses this value to adjust the spray gun dwell time, the workpiece's residence time in the blasting zone, or to adjust the conveyor speed and convert it into an equivalent processing time. Pressure channel allocation coefficient. The structure originates from the gain scheduling concept in control theory, which dynamically allocates the action ratio of each control channel according to the current error state. Based on this, and combined with the process experience in pure titanium blasting that "the side with stronger surface action is more suitable for prioritizing pressure reduction, while the side with weaker surface action is more suitable for increasing processing time compensation," the following relationship is formed: ; in, Indicates the pressure channel allocation coefficient; This represents the control deviation value output from step two; This represents the directional distribution gain, the value of which comes from the pure titanium test spray calibration results during the equipment commissioning phase and is stored in the control unit parameter table. Its value satisfies... ; This represents the smoothing coefficient, determined during the debugging phase based on the stability adjustment requirements within the target neighborhood, and is taken as a positive value. The derivation logic of this formula is as follows: First, using... As the basic equal proportion of the two channels of pressure and time, we then introduce... As a continuous direction-dependent modulation term. This modulation term originates from a saturated normalization structure with an absolute value denominator, when When this value is positive, it indicates that the current surface effect is higher than the target surface standard, and the pressure channel weight increases with the increase of deviation; when... When this item is negative, it indicates that the current surface effect is lower than the target surface standard, and the pressure channel weight is lower than the basic average proportion, thus allocating more correction space to the processing time channel for detailed compensation. Because ,and ,therefore It always falls between 0 and 1, and can be directly used as the allocation coefficient for subsequent dual-channel correction calculations.

[0031] After obtaining the pressure channel allocation coefficient Then, a dual-channel correction relationship is established based on the classical proportional control concept. Its original source is the proportional adjustment formula in control theory, which states that the control quantity changes proportionally to the error. The derivation made in this step extends single-channel proportional control to "dual-channel proportional control after directional allocation," ensuring that the same control deviation value... Capable of simultaneously generating sandblasting pressure adjustment amount and processing time adjustment amount Furthermore, the ratio of the effects of the two channels is determined by the formula in the previous equation. Automatic allocation. The specific relationships are as follows: ; in, Indicates the sandblasting pressure adjustment amount; Indicates the amount of processing time adjustment; The pressure channel proportionality coefficient is obtained by calibration through multiple sets of pure titanium test spraying experiments during the equipment commissioning phase, and is used to establish the proportional relationship between the control deviation value and the pressure correction amount. This represents the time channel proportionality coefficient, which is obtained from the same batch of debugging experiments and is used to establish the proportional relationship between the control deviation value and the processing time correction amount. This represents the pressure channel allocation coefficient calculated from the previous formula; This represents the control deviation value output in step two. There is a clear logical relationship between this formula and the previous one: the previous formula first gives the proportion of the pressure channel in the overall regulation under the current deviation state, and the subsequent formula decomposes the control deviation value into two executable correction values ​​based on this proportion. The negative sign in the formula is used to ensure that the correction direction is always opposite to the deviation direction, i.e., when... When the current surface effect is higher than the target value, the system calculates... and Negative values ​​correspond to reducing sandblasting pressure and shortening processing time, respectively; when At that time, the system calculated and A positive value corresponds to increasing the sandblasting pressure and extending the processing time, respectively. Because... and Because they are complementary, the same control deviation value is conserved and distributed between the pressure channel and the time channel. This structure allows coarse and fine adjustments to switch naturally within a continuous formula without the need for additional control mode switching.

[0032] Substituting the above process into specific operating conditions yields a complete calculation implementation example. Let the control deviation output in step two be... This indicates that the current surface finish is higher than the target surface standard, and the sandblasting effect needs to be reduced. Further calibration during the debugging phase yields the directional distribution gain. Assigning smoothing coefficients Pressure channel proportionality coefficient Time channel scaling factor First, calculate the pressure channel distribution coefficient using the first formula: ,but ,have Multiply by get ,thereby Substituting this result into the second equation, we get... , This result indicates that when the surface impact is too strong, the system allocates a larger portion of the correction to the blasting pressure channel, causing the blasting pressure to be preferentially reduced. Simultaneously, this is combined with a slight reduction in processing time to quickly weaken the surface impact. Let's assume that the control deviation output of step two at another time is... This indicates that the current surface finish is slightly below the target surface standard, requiring enhanced sandblasting. , ,have Multiply by Get about ,thereby Substituting this into the second equation, we get... , This result indicates that in the weaker region near the target, the system allocates a larger share of correction to the processing time channel, while simultaneously using a smaller amount of sandblasting pressure compensation, allowing the pure titanium surface to maintain smooth convergence as it approaches the target sandy feel. The control unit then... Write the valve control setpoint correction register, Write the sandblasting time or conveyor speed correction register to form an adaptive correction amount for process parameters that can be directly executed in the next step.

[0033] S4: The sandblasting pressure adjustment amount and the processing time adjustment amount are respectively added to the current sandblasting pressure setting value and the current processing time setting value to generate updated sandblasting pressure setting value and updated processing time setting value, and the sandblasting machine actuator is driven to perform sandblasting treatment according to the updated sandblasting pressure setting value and the updated processing time setting value.

[0034] Specifically, step four directly receives the sandblasting pressure adjustment amount output from step three. and processing time adjustment amount These two corrections are then implemented in the sandblasting machine control unit, ensuring that the current workpiece or processing section completes the sandblasting process according to the updated process parameters. Simultaneously, the executed process state is naturally transitioned to the next round of surface effect estimation. In practice, the sandblasting machine control unit continuously stores the current sandblasting pressure setting. and the current processing time setting value ,in It is the target pressure value written to the pressure regulating execution module at the end of the previous control cycle. This is the target processing duration value written to the timing execution module at the end of the previous control cycle. (Step 3 output) and Upon reaching the control unit, instead of directly replacing the original setpoint, the new setpoint is incrementally updated and superimposed onto the current setpoint to form the new setpoint for the next control cycle. This originates from the incremental execution model in discrete control theory, where "the new setpoint equals the current setpoint plus the correction increment." Building upon this, this step incorporates the physical response characteristics of the sandblasting equipment's execution end, introducing an execution smoothing coefficient before the increment to ensure continuous transition between valve-controlled and timing-based execution when parameters change. The corresponding relationship is written as: ; in, This indicates the updated sandblasting pressure setting value, which will be used for the sandblasting treatment section corresponding to the next sampling cycle and written to the pressure regulation execution module; This indicates the updated processing time setting value, which will be used for the sandblasting processing segment corresponding to the next sampling cycle and written into the sandblasting timing execution module; This indicates the current sandblasting pressure setting value, which is saved by the control unit in the previous control cycle; This indicates the current processing time setting, which is saved by the control unit in the previous control cycle; This indicates the sandblasting pressure adjustment amount output in step three. This indicates the adjustment amount for the processing time output in step three; This represents the smoothing coefficient for the pressure channel. Its value is determined during equipment commissioning based on the response curve of the pressure regulating valve and the stabilization time of the spray gun outlet pressure, and it must satisfy the following conditions: ; This represents the smoothing coefficient for the time channel execution. Its value is determined during the equipment commissioning phase based on the sandblasting cycle time, the response time of the conveying mechanism, and the workpiece passing rhythm, and must satisfy the following conditions: The two lines of this formula have a parallel but related logical relationship: the first line updates the pressure channel setpoint, and the second line updates the time channel setpoint; together, they constitute the execution benchmark for the sandblasting treatment section corresponding to the next sampling cycle. Because... , The coefficient is dimensionless. and , and Since they belong to the same category of quantities, the left and right sides of the two equations remain consistent. Using this notation, the adaptive correction values ​​for the process parameters given in step three are converted into new setpoints in the actual controller.

[0035] get and Subsequently, the control unit drives the pressure execution link and the time execution link respectively to complete the sandblasting action. The pressure execution link typically consists of the controller's analog-to-digital output module, power drive module, and electromagnetic pressure regulating valve or proportional valve. The controller will... The corresponding valve control electrical signal is converted and written into the output register. The power drive module outputs a control current based on this signal. The pressure regulating valve changes the valve core opening of the air supply branch according to this current, thereby adjusting the target pressure in the front chamber of the spray gun. After the new pressure setting is reached, the blasting medium acts on the pure titanium surface with an updated blasting intensity. The time execution link typically consists of a blasting timing module and a transmission control module. The controller will... After the sandblasting duration register is written, the gun dwell logic or the conveyor speed control logic executes according to the set value, synchronously correcting the duration of the sandblasting medium's action on the workpiece surface. Since pure titanium surfaces are highly sensitive to both blasting intensity and action time, the coordinated updating of the two channels within the same control cycle ensures that the "pressure priority or time priority" correction strategy obtained in step three is truly implemented in the equipment operation. After execution, the updated process state enters the next sampling cycle. The sandblasting pressure is re-acquired by the pressure detection element, the processing time is re-accumulated by the control program, and the ambient humidity continues to be collected by the humidity detection element. These three data points together form the input basis for the next round of artificial intelligence surface effect estimation model. To ensure that the executed process state can stably enter the next round of estimation, this step sets the sequence of "parameter update—sandblasting execution—state resampling" in the control program, i.e., first writing... and Then, after completing an actual sandblasting section, the next sampling cycle is triggered, thus ensuring that the next step in step one uses the process state that has actually been applied to the workpiece surface, rather than just the theoretical state that remains at the set layer.

[0036] A specific example can more clearly illustrate the execution process. Suppose that in a certain control cycle, step three outputs... , The sandblasting pressure setting currently stored in the control unit is The current processing time setting is During the equipment commissioning phase, the pressure channel smoothing coefficient is determined. Time channel execution smoothing coefficient Substituting into the above equation, we get... , The control unit will then... The corresponding valve control value is written into the pressure regulating execution module, and... The timing execution module is written to ensure that the sandblasting machine performs sandblasting with a lower blasting intensity and shorter action time in the sandblasting treatment section corresponding to the next sampling cycle. For example, in another control cycle, step three outputs... , If the current setting value is still taken , Then there is , This result indicates that when the surface effect is slightly below the target value, the system compensates by slightly increasing the pressure and compensating with a slightly longer processing time. After executing this set of new settings, the pressure sensing element will read the new pressure state at the next sampling moment, the timing module will record the new sandblasting duration, and the humidity sensing element will continue to provide the current ambient humidity. This data is re-input into the artificial intelligence surface effect estimation model in step one to generate a new surface effect estimate. This leads to the next round of continuous control process: "surface estimation - deviation generation - parameter correction - execution update".

[0037] In one or more embodiments, such as Figure 2 As shown, an intelligent control system for a sandblasting machine based on adaptive process parameters is disclosed. The system includes: The effect estimation module is used to collect the current sandblasting pressure, processing time, and ambient humidity of the sandblasting machine; input the pre-deployed artificial intelligence surface effect estimation model, and generate the current surface effect estimate value based on the data collected by the process parameter acquisition module; The deviation generation module is used to store the target surface standard value and generate a control deviation value based on the current surface effect estimate and the target surface standard value. The parameter correction module is used to calculate the pressure channel allocation coefficient based on the control deviation value, and generate the sandblasting pressure adjustment amount and the processing time adjustment amount. The execution control module is used to add the sandblasting pressure adjustment amount and the processing time adjustment amount to the current set value to generate an updated set value, and drive the sandblasting machine to perform sandblasting processing according to the updated set value.

[0038] It is worth noting that the specific workflow of the intelligent control system for sandblasting machines based on adaptive process parameters provided in this embodiment of the invention is the same as that of the intelligent control method for sandblasting machines based on adaptive process parameters described in the above embodiments, and will not be repeated here.

[0039] This invention also provides an intelligent control device for a sandblasting machine based on adaptive process parameters, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the steps described in the above embodiments of the intelligent control method for a sandblasting machine based on adaptive process parameters. Figure 1 The steps S1 to S4 described above; or, when the processor executes the computer program, it implements the functions of each module in the above system embodiments.

[0040] For example, the computer program may be divided into one or more modules, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules may be a series of computer program instruction segments capable of performing specific functions, which describe the execution process of the computer program in the intelligent control device for the sandblasting machine based on adaptive process parameters.

[0041] The intelligent control device for a sandblasting machine based on adaptive process parameters can be a desktop computer, laptop, handheld computer, or cloud server, etc. This intelligent control device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the intelligent control device may also include input / output devices, network access devices, buses, etc.

[0042] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the process parameter adaptive intelligent control equipment for sandblasting machines, connecting all parts of the equipment via various interfaces and lines.

[0043] The memory can be used to store the computer program and / or modules. The processor realizes various functions of the intelligent control equipment for the sandblasting machine based on process parameter adaptation by running or executing the computer program and / or modules stored in the memory and calling the data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function, etc.; the data storage area may store data created according to the operation of the controller, etc. In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital card (SD card), flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0044] The integrated module of the sandblasting machine intelligent control equipment based on adaptive process parameters, if implemented as a software functional unit and sold or used as an independent product, can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0045] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0046] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A sandblasting machine intelligent control method based on adaptive process parameters, characterized in that, The method includes: The current sandblasting pressure, processing time, and ambient humidity of the sandblasting machine are collected and input into the artificial intelligence surface effect estimation model pre-deployed in the control unit to generate the current surface effect estimate. Obtain the pre-stored target surface standard value, compare the current surface effect estimate with the target surface standard value, and generate a control deviation value that includes deviation direction, deviation degree, and directional enhancement information; Calculate the pressure channel allocation coefficient based on the control deviation value, and allocate the control deviation value as sandblasting pressure adjustment amount and processing time adjustment amount according to the pressure channel allocation coefficient; The sandblasting pressure adjustment and the processing time adjustment are respectively added to the current sandblasting pressure setting and the current processing time setting to generate updated sandblasting pressure setting and updated processing time setting, and the sandblasting machine actuator is driven to perform sandblasting according to the updated sandblasting pressure setting and the updated processing time setting.

2. The intelligent control method for sandblasting machines based on adaptive process parameters according to claim 1, characterized in that, The generation of the current surface effect estimate specifically includes: The collected sandblasting pressure, processing time, and ambient humidity are normalized, and the normalized values ​​are used to form an input vector to input the artificial intelligence surface effect estimation model. The artificial intelligence surface effect estimation model is a feedforward neural network containing an input layer, two hidden layers and an output layer. The single-value result output by the output layer is the current surface effect estimate.

3. The intelligent control method for sandblasting machines based on adaptive process parameters according to claim 1, characterized in that, The step of comparing the current surface effect estimate with the target surface standard value to generate a control deviation value containing information on deviation direction, deviation degree, and directional enhancement specifically includes: The original surface deviation is obtained by subtracting the target surface standard value from the current surface effect estimate; The original surface deviation is introduced with a target neighborhood smoothing term and a directionality enhancement term to generate the control deviation value after nonlinear shaping. Specifically, when the original surface deviation is greater than zero, the amplitude of the control deviation increases as the deviation increases; when the original surface deviation is less than zero, no directionality enhancement is performed.

4. The intelligent control method for sandblasting machines based on adaptive process parameters according to claim 1, characterized in that, The calculation of the pressure channel allocation coefficient based on the control deviation value specifically includes: The control deviation value is processed using a saturated normalization function that includes directional allocation gain and allocation smoothing coefficient, so that the pressure channel allocation coefficient falls within the range of 0 to 1. Specifically, when the control deviation value is greater than zero, the pressure channel allocation coefficient is greater than the basic average proportion; when the control deviation value is less than zero, the pressure channel allocation coefficient is less than the basic average proportion.

5. The intelligent control method for a sandblasting machine based on adaptive process parameters according to claim 1, characterized in that, The step of allocating the control deviation value into sandblasting pressure adjustment and processing time adjustment based on the pressure channel allocation coefficient specifically includes: The sandblasting pressure adjustment amount is obtained by multiplying the product of the pressure channel allocation coefficient and the pressure channel proportional coefficient and the control deviation value. The processing time adjustment amount is obtained by multiplying the product of the pressure channel allocation coefficient and the time channel proportional coefficient and the control deviation value. The directions of the sandblasting pressure adjustment and the processing time adjustment are opposite to the direction of the control deviation value.

6. The intelligent control method for a sandblasting machine based on adaptive process parameters according to claim 1, characterized in that, The step of adding the sandblasting pressure adjustment amount and the processing time adjustment amount to the current sandblasting pressure setting value and the current processing time setting value, respectively, is as follows: The sandblasting pressure adjustment amount is multiplied by the pressure channel smoothing coefficient and then added to the current sandblasting pressure setting value to generate the updated sandblasting pressure setting value. The updated processing time setting is generated by multiplying the processing time adjustment by the time channel smoothing coefficient and adding it to the current processing time setting.

7. The intelligent control method for a sandblasting machine based on adaptive process parameters according to claim 1, characterized in that, The drive mechanism of the sandblasting machine performs sandblasting according to the updated sandblasting pressure setting and the updated processing time setting, specifically including: The updated sandblasting pressure setting is converted into a control current signal to drive the electromagnetic pressure regulating valve or proportional valve to adjust the valve core opening of the air supply branch. Write the updated processing time setting into the sandblasting timing execution module to adjust the spray gun dwell time or the conveyor speed.

8. The intelligent control method for a sandblasting machine based on adaptive process parameters according to claim 1, characterized in that, The training method for the artificial intelligence surface effect estimation model is as follows: Pure titanium samples with consistent initial surface conditions were selected. Under the condition of keeping the spray gun structure, spray distance and sand specifications consistent, the combination of sandblasting pressure, processing time and ambient humidity was changed to form multiple working conditions. After sandblasting under each working condition, the surface of the sample is tested offline and a comprehensive surface evaluation value is generated. The neural network is trained under supervision using the multiple sets of working conditions as input samples and the comprehensive surface evaluation value as the output label.

9. The intelligent control method for a sandblasting machine based on adaptive process parameters according to claim 1, characterized in that, The acquisition of the pre-stored target surface standard value specifically includes: Pure titanium standard samples that meet appearance requirements were prepared using stable sandblasting conditions; The pure titanium standard sample was subjected to offline testing, and the offline testing results were converted into single-value surface evaluation results according to the evaluation rules consistent with those used when training the artificial intelligence surface effect estimation model.

10. An intelligent control system for a sandblasting machine based on adaptive process parameters, characterized in that: include: The effect estimation module is used to collect the current sandblasting pressure, processing time, and ambient humidity of the sandblasting machine; input the pre-deployed artificial intelligence surface effect estimation model, and generate the current surface effect estimate value based on the data from the process parameter acquisition module; The deviation generation module is used to store the target surface standard value and generate a control deviation value based on the current surface effect estimate and the target surface standard value. The parameter correction module is used to calculate the pressure channel allocation coefficient based on the control deviation value, and generate the sandblasting pressure adjustment amount and the processing time adjustment amount. The execution control module is used to add the sandblasting pressure adjustment amount and the processing time adjustment amount to the current set value to generate an updated set value, and drive the sandblasting machine to perform sandblasting processing according to the updated set value.

Images