Computer control method and system for flotation production process

Abstract

The invention discloses a computer control method and system for a flotation production process, and relates to the technical field of computer flotation control, and the method comprises the steps: collecting multi-modal data, including graphic frame data and physical quantity vectors, in a flotation tank, calculating a foam reflectivity to determine a light source brightness color temperature adjustment coefficient, carrying out enhanced correction, and constructing an image super-division model; outputting a high-resolution physical consistent reconstruction image; and calculating the unit mass of the reconstructed graph, calculating the collision force between the partitions, carrying out time advance and dynamic evolution on all segmented regions, and carrying out consistency error verification. According to the method, the neighborhood pixels are enhanced and corrected by combining the calibrated defect correction parameters, so that the detail level and the signal-to-noise ratio of the image are improved, systematic errors caused by equipment are corrected, the quality of the super-resolution input image is higher, the spatial detail information of the image is richer, and the physical consistency is higher.

Description

Technical Field

[0001] The present invention relates to the technical field of computer flotation control, in particular to a computer control method and system for a flotation production process. Background Art

[0002] Flotation technology is an important method in the field of mineral processing, used to separate mineral particles from gangue particles. Its basic principle is to allow bubbles to adhere to mineral particles and make them float to the liquid surface for separation, while suppressing the gangue particles to achieve separation. As the resource industry places higher demands on flotation process efficiency and product quality, automation, intelligence and precise control of the flotation process have become key development directions. Currently, computer control technology for the flotation process is gradually being introduced to monitor the state of the foam layer, changes in physical parameters and dynamic fluid behavior in real time, and to control the process based on this data.

[0003] However, existing flotation control systems usually still rely on the acquisition and analysis of a single type of data to establish a comprehensive information interaction model, and are unable to fully understand the flotation state from multiple levels of enhanced images, thermal parameters and dynamic fluid behavior. Secondly, there is little research on foam microdynamics in existing technologies. Although some flotation control technologies have introduced computational models based on foam distribution or foam size, such models often fail to effectively capture complex dynamic evolution characteristics, and therefore have limited ability to describe partitioned collision forces and time evolution response mechanisms. In addition, existing flotation automation equipment is still not perfect in data anomaly monitoring, and lacks the ability to analyze abnormal cases based on the prediction of main state vectors, physical quantity parameters and change rates, making it difficult to achieve long-term stability prediction and optimization of the flotation process. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a computer control method for a flotation production process to solve the problem that existing flotation control systems generally rely on the acquisition and analysis of a single type of data to establish a comprehensive information interaction model, and are unable to fully understand the flotation state from multiple levels of enhanced images, thermal parameters and dynamic fluid behavior. Secondly, there is little research on foam microdynamics in the existing technology. Although some flotation control technologies have introduced calculation models based on foam distribution or foam size, such models often fail to effectively capture complex dynamic evolution characteristics, and therefore have limited ability to describe partitioned collision forces and time evolution response mechanisms. In addition, existing flotation automation equipment is not yet perfect in data anomaly monitoring, and lacks the ability to analyze abnormal cases based on the prediction of main state vectors, physical quantity parameters and change rates, making it difficult to achieve long-term stability prediction and optimization of the flotation process.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a computer control method for a flotation production process, comprising:

[0008] The system collects multimodal data in the flotation tank, including graphic frame data and physical quantity vectors, calculates the foam reflectivity to determine the light source brightness and color temperature adjustment coefficient, performs enhancement correction, and constructs an image super-resolution model to output a high-resolution physically consistent reconstruction image.

[0009] Calculate the unit mass of the reconstructed image and the collision force between partitions. Perform time advancement and dynamic evolution on all partitioned regions and verify the consistency error.

[0010] Determine the Reynolds number, foam volume fraction, foam coverage, and surface tension coefficient combination as the main state vector. Construct a derivative equation for the rate of change based on the main state vector, physical quantity vector, and disturbance term. Predict the rate of change and generate a contribution heat map. Combined with the consistency error verification results, perform proportional-integral (PI) control adjustments on the term with the largest contribution. After the adjustments, re-verify the consistency error.

[0011] Record operation data to build a historical abnormal case library.

[0012] As a preferred solution of the computer control method for the flotation production process of the present invention, wherein: the output high-resolution physical consistent reconstruction map includes:

[0013] Use a light meter to read the light intensity on the foam surface, and use the collected graphic frame data to measure the foam grayscale at the center of the foam area, and determine the grayscale mean of the actual grayscale area of ​​the foam to determine the foam reflectivity;

[0014] Based on the standard light intensity reference value and standard color temperature reference value of the foam area determined by historical experience, the multi-spectral lighting compensation coefficient is calculated, including the adjustment coefficients of brightness and color temperature;

[0015] Adjustment coefficients based on brightness and color temperature are input and adjusted through the LED light power control to complete dimming and color adjustment, and light correction graphic frame data is re-collected to perform enhanced correction based on field pixels;

[0016] The watershed algorithm is used to segment the enhanced graphic frame data to obtain different independent foam partitions at the same time;

[0017] The enhanced image frame data is aligned with the normalized physical quantity vectors for timestamp data, which is then fed into the image super-resolution model. The model is then trained on historical enhanced images and high-resolution annotated images, where each sample is labeled with a physical parameter.

[0018] Output a high-resolution physically consistent reconstruction image based on the trained image super-resolution model;

[0019] The watershed algorithm is used again to segment the high-resolution physically consistent reconstruction image. Within the i-th segmented subdomain after re-segmentation, the two point pairs with the largest Euclidean distance are found, and the pixel Euclidean distance of the point pair coordinates is determined as the maximum main path of the reconstructed image partition.

[0020] The overall distribution mean is calculated based on the major axis lengths of the reconstructed image partitions and the total number of segmented regions.

[0021] As a preferred solution of the computer control method for the flotation production process of the present invention, the calculation of the unit mass of the reconstruction map and the collision force between partitions are performed, and time advancement and dynamic evolution of all partitioned areas are performed, and consistency error verification is performed, including:

[0022] According to the normalized physical quantity vector, the simulation environment is built, the simulation space size is determined according to the high-resolution physical consistent reconstruction map, and the unit mass is preliminarily calculated according to the standard value of foam density;

[0023] Synchronously calculate the fluid drag force by combining the solution dynamic viscosity and the main direction velocity of the fluid, and calculate the collision force between partitions;

[0024] Determine the surface tension based on the interface length of the segmented region, the surface tension coefficient, and the normal direction outside the segmented region, and perform time advancement and dynamic evolution on all segmented regions.

[0025] Grid mapping is performed based on the position information and maximum main diameter of each reconstructed image partition after advancement, and a simulation image is generated. The consistency error is verified with the high-resolution physical consistent reconstruction image.

[0026] The sum of the historical mean and standard deviation is used as the difference threshold. If the average pixel difference value is less than the difference threshold, it means that the error is small.

[0027] As a preferred embodiment of the computer control method for the flotation production process of the present invention, the method comprises: determining the Reynolds number, foam volume fraction, foam coverage, and surface tension coefficient as a main state vector, constructing a derivative equation of the change rate based on the main state vector, the physical quantity vector, and the disturbance term, and predicting the change rate, including:

[0028] The ratio of the total area of ​​the segmented region to the total area of ​​the image is used as the foam volume fraction, and the Reynolds number is calculated based on the solution flow rate, solution density and solution dynamic viscosity;

[0029] The foam coverage is calculated by the ratio of the number of pixels in the foam area to the total number of pixels in the super-resolution image, and the foam volume fraction, Reynolds number, foam coverage and surface tension coefficient are combined into the main state vector;

[0030] The influence coefficients of the main process parameters on the state are determined based on the main state vector, physical quantity vector, and temperature data at the corresponding time. At the same time, the image mechanics field coupling differential terms are determined based on the gradient field of the reconstructed image partition and the main state vector. Combined with the disturbance term, the derivative equation of the main state vector with respect to time is constructed to represent the rate of change of the system state.

[0031] Use historical data and discrete difference approximation to calculate the rate of change of the main state;

[0032] Construct data tables based on the main state vector, physical quantity vector, temperature, image gradient, and change rate, where each row represents data at a specific moment.

[0033] The fourth-order Runge-Kutta method RK4 is used for time advancement to update the main state vector, physical quantity vector, temperature, image gradient and predict the rate of change based on the derivative equation.

[0034] As a preferred solution of the computer control method for the flotation production process of the present invention, wherein: generating the contribution heat map, combining the consistency error verification result, performing proportional integral PI control adjustment on the maximum contribution item, and re-performing consistency error verification after adjustment, including:

[0035] According to the change rate calculation of the derivative equation, the influence of the main process parameters on the state, the image mechanics field coupling differential term and the disturbance term on the main state change rate are normalized and calculated as the contribution;

[0036] According to the main state vector and contribution value, a heat map data table is organized and a heat map is drawn using a drawing tool. The sum of the historical mean and the standard deviation is used as the contribution threshold for each contribution. If the contribution of the data item is greater than or equal to the contribution threshold, it is determined that the data item needs to be adjusted, and proportional integral PI control is used for adjustment. After adjustment, the consistency error verification is performed again.

[0037] As a preferred solution of the computer control method for the flotation production process of the present invention, wherein: the multimodal data collected in the flotation cell includes:

[0038] Multi-angle, high-frame-rate industrial CCD+spectral enhancement cameras are deployed simultaneously at key locations in the flotation tank to acquire multimodal data on the 3D morphology and texture of the foam surface, as well as process data such as flow rate, pH, conductivity, bubble size distribution, and reagent concentration, as well as graphic frame data.

[0039] The collected real-time flow rate of flotation cell feed, pH value of flotation cell slurry, slurry conductivity, average diameter of foam in the foam zone, and online concentration of reagent are preprocessed and timestamped to form a physical quantity vector.

[0040] As a preferred solution of the computer control method for the flotation production process of the present invention, wherein: the recording operation data to construct a historical abnormal case library includes:

[0041] If the average pixel difference value is greater than or equal to the difference threshold, it means that the error is large, and the main state vector, physical quantity vector, and temperature data are recorded as abnormal data to build a historical abnormal case library.

[0042] In a second aspect, the present invention provides a computer control system for a flotation production process, comprising:

[0043] Data processing module, which obtains multimodal data of the flotation cell through sensors and acquisition equipment;

[0044] Image reconstruction module, builds an image super-resolution model, generates a high-resolution physically consistent reconstruction image, and calculates the maximum main diameter of the segmented area and the overall distribution mean;

[0045] The evolution simulation module promotes the dynamic evolution of all segmented areas in the time dimension, generates simulated images of foam behavior, and uses the simulated images to verify consistency errors;

[0046] The rate prediction module calculates fluid dynamics related parameters, generates the main state vector, constructs the derivative equation of the change rate, and predicts the change rate of foam and material flow;

[0047] The contribution monitoring module calculates the impact of the main process parameters on the system, generates a heat map of the impact contribution, and identifies anomalies based on the results of consistency error verification.

[0048] In a third aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, any step of the computer control method for the flotation production process as described in the first aspect of the present invention is implemented.

[0049] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the computer control method for a flotation production process as described in the first aspect of the present invention is implemented.

[0050] The beneficial effects of the present invention are as follows: by combining calibrated defect correction parameters to enhance and correct neighborhood pixels, not only the image detail level and signal-to-noise ratio are improved, but also the systematic errors caused by the equipment itself are corrected, so that the super-resolution input image quality is higher, the spatial detail information of the image is richer and the physical consistency is stronger, and the watershed algorithm is used to segment the foam area of ​​the enhanced image, which can effectively identify independent foam structures that are not connected to each other in the image, and the alignment of the timestamp and the physical quantity vector enables the physical parameters to be accurately mapped to the specific foam area, ensuring the authenticity and supervision effect of the training samples, and by calculating the collision force between the segmented areas and simulating the elastic interaction within the foam group, the collision and extrusion phenomena between the foams are effectively captured, so that the foam movement and deformation during the simulation process conform to the complex collective dynamic characteristics of the foam medium, and by constructing the rate of change derivative equation, it not only reflects the dynamic influence of the main state on itself, but also introduces multi-scale coupling and external disturbances, so that the model can respond to real-time changes in environmental and process conditions, and fully reflect the complex nonlinear dynamic characteristics of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0052] Figure 1 Schematic diagram of the flow chart of the computer control method for the flotation production process in Example 1.

[0053] Figure 2 Schematic diagram of the structure of the computer control system for the flotation production process in Example 1. DETAILED DESCRIPTION

[0054] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0055] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0056] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0057] Example 1, with reference to Figures 1 to 2 , which is the first embodiment of the present invention, provides a computer control method for a flotation production process, comprising the following steps:

[0058] S1, located in the flotation tank, collects multimodal data, including graphic frame data and physical quantity vectors, calculates foam reflectivity to determine the light source brightness and color temperature adjustment coefficient, performs enhancement correction, and builds an image super-resolution model to output a high-resolution physically consistent reconstruction image;

[0059] Preferably, the multimodal data collected in the flotation cell includes:

[0060] Multi-angle, high-frame-rate industrial CCD and spectral enhancement cameras are deployed simultaneously at key locations in the flotation tank to acquire multimodal data on the 3D morphology and texture of the foam surface. These cameras also collect flow, pH, conductivity, bubble size distribution (laser particle size analyzer), reagent concentration process data, and graphic frame data. (All devices are synchronized with a unified high-precision clock (using the PTP / IEEE1588 clock protocol).

[0061] The collected real-time flow rate of flotation cell feed, pH value of flotation cell slurry, slurry conductivity, average diameter of foam in the foam zone, and online concentration of reagent are preprocessed, time-stamped, and composed into physical quantity vectors.

[0062] Through the combination of a multi-angle, high-frame-rate industrial CCD and a spectrally enhanced camera, the three-dimensional morphology and texture details of the foam surface can be captured, achieving a multi-dimensional characterization of the spatial distribution of the foam structure and the interface characteristics, providing rich and high-quality visual information on the foam quality and dynamic changes. The application of real-time synchronous acquisition and a unified clock protocol ensures the accurate time alignment of key process parameters such as flow, pH, conductivity, bubble particle size and reagent concentration with the image frame data, eliminating the time deviation between multi-source data and providing a solid spatiotemporal foundation for multimodal data fusion. After preprocessing and data fusion, the physical quantity vector formed not only systematically reflects the fluid chemical environment, foam physical properties and process dosing conditions in the flotation tank.

[0063] Furthermore, the output is a high-resolution physically consistent reconstruction map, including:

[0064] Use a light meter to read the light intensity on the foam surface, and use the collected graphic frame data to measure the foam grayscale at the center of the foam area. Then determine the grayscale mean of the actual grayscale area of ​​the foam, calculate the ratio of the grayscale mean to the foam grayscale, and determine the foam reflectivity.

[0065] The multi-spectral lighting compensation coefficients, including the adjustment coefficients of brightness and color temperature, are calculated based on the standard light intensity reference value and standard color temperature reference value of the foam area determined based on historical experience, and are expressed as:

[0066]

[0067] Among them S LES Indicates the adjustment coefficient of LED light source brightness, S CT Indicates the adjustment coefficient of LED light source color temperature, L ref Indicates the standard light intensity reference value of the foam area, L t Indicates the light intensity of the foam surface, Indicates the standard color temperature reference value, R t Indicates the foam reflectivity;

[0068] The adjustment coefficient based on brightness and color temperature is input and adjusted through the LED light power supply control to complete the dimming and color adjustment, and the light correction graphic frame data is re-collected and enhanced correction is performed according to the field pixels, which is expressed as:

[0069]

[0070] in Represents the pixel value of the enhanced image at position (x, y) and band λ, N(x, y) represents the number of neighborhood pixels centered at (x, y), represents the neighborhood index set around the pixel (x, y), α and β represent the defect correction parameters of the device factory calibration, which can be determined by looking up the table in the LED and camera manuals. t (i, j, λ) represents the pixel value of the original image at position (x, y) and band λ;

[0071] The watershed algorithm is used to segment the enhanced graphic frame data to obtain different independent foam partitions at the same time;

[0072] The enhanced image frame data is aligned with the normalized physical quantity vector for timestamp data, input into the image super-resolution model, and trained together with the historical enhanced images and high-resolution annotated images. Each sample is labeled with a physical parameter. The training loss is calculated as:

[0073]

[0074] D py =∑ i (Si,t -D b,t ) 2 ;

[0075] Among them L sup represents the total loss of the super-resolution reconstruction network, I HR,t represents the experimental calibration high-resolution foam image at time t, represents the physical constraint high-resolution foam image output by the image super-resolution model at time t, λ py represents the physical loss weight factor, Represents the normalized physical quantity vector as well as Physical consistency loss function, S i,t represents the ith independent foam partition at time t, D b,t It represents the average physical particle size of the foam at time t, which can be directly measured by an online particle size analyzer;

[0076] The trained image super-resolution model outputs a high-resolution physically consistent reconstruction map. The resulting high-resolution physically consistent reconstruction map is a super-resolution structural map with significantly enhanced spatial details and physical consistency. It can be used in subsequent processes such as arbitrary morphological dynamics inversion, local foam analysis, and physical field parameter tracing. Physical constraints ensure structural reliability.

[0077] The watershed algorithm is used again to segment the high-resolution physical consistent reconstruction image. In the i-th segmentation subdomain after re-segmentation, the two point pairs with the largest Euclidean distance are found, and the pixel Euclidean distance of the point pair coordinates is determined as the maximum main path of the reconstructed image partition, which is expressed as:

[0078]

[0079] Where (x a ,y a ),(x b ,y b )∈sub i represents the pixel pair belonging to the i-th re-segmented foam subdomain, where (x a ,y a ),(x b ,y b ) are the coordinates of two pixel points respectively, represents the maximum main path of the i-th reconstructed image partition at time t;

[0080] The overall distribution mean is calculated based on the major axis lengths of the reconstructed image partitions and the total number of segmented regions.

[0081] By using a light meter to obtain the actual light intensity on the foam surface and combining it with the grayscale center and local area mean of the image, the foam reflectivity is accurately calculated. This can effectively correct the grayscale deviation in the foam image caused by differences in reflective properties, thereby enhancing the accuracy of the image in reflecting the actual physical state of the foam.

[0082] Based on the reference values ​​of the standard light intensity and color temperature of the historical foam area, the brightness and color temperature adjustment coefficients are calculated to achieve dynamic power control and adjustment of the LED light source, so that the collected ambient light is optimally matched, the image brightness and color restoration are closer to reality, and the impact of ambient light changes on image quality and subsequent analysis accuracy is effectively reduced. The neighborhood pixels are enhanced and corrected in combination with the calibrated defect correction parameters, which not only improves the image detail level and signal-to-noise ratio, but also corrects the systematic errors caused by the equipment itself, making the super-resolution input image higher in quality, with richer spatial detail information and stronger physical consistency. The watershed algorithm is used to segment the enhanced image into foam areas, which can effectively identify independent foam structures that are not connected to each other in the image. The combination of timestamps and physical quantity vector alignment enables the physical parameters to be accurately mapped to specific foam areas, ensuring the authenticity of the training samples and the supervision effect. Therefore, the dual accuracy of spatial resolution and physical properties of the reconstructed image is improved through super-resolution model training with physical constraints.

[0083] During the training process, a physical consistency loss term is introduced. Combined with the foam particle size labels measured by the real particle size analyzer, the super-resolution model not only improves the image vision, but also ensures a high degree of consistency between the structure and physical properties, enhancing the model's generalization and real perception capabilities for complex foam scenes. The high-resolution reconstructed image is segmented again to extract the maximum principal diameter, and the main scale of the foam segmentation area is reflected by calculating the maximum Euclidean distance pixel pair, accurately depicting the spatial characteristics of each foam structure. The overall distribution mean calculated based on the main scale and quantity of all segmented areas provides reliable statistical characteristic indicators for flotation process parameters and foam dynamic changes, which can effectively assist process optimization decisions and realize real-time status monitoring.

[0084] S2, calculate the unit quality of the reconstructed image, calculate the collision force between partitions, perform time advancement and dynamic evolution on all partitioned regions, and perform consistency error verification;

[0085] Preferably, the unit quality of the reconstructed image is calculated, and the collision force between partitions is calculated. Time advancement and dynamic evolution are performed on all partitioned regions, and consistency error verification is performed, including,

[0086] According to the normalized physical quantity vector, the simulation environment is built. The simulation space size is determined according to the high-resolution physical consistent reconstruction map. The unit mass is preliminarily calculated according to the standard value of foam density, which is expressed as:

[0087]

[0088] where m i represents the quality of the ith simulation unit, d i represents the maximum main path of the i-th reconstructed image partition, p f Indicates foam density;

[0089] The fluid drag force is calculated by combining the dynamic viscosity of the solution and the velocity in the main direction of the fluid, and the collision force between partitions is calculated, which is expressed as:

[0090] F fl，i =3πηd i (uv i );

[0091]

[0092] Among them F fl，i represents the fluid drag force of the i-th reconstruction image partition, F col，ij represents the collision force of the i-th and j-th reconstruction image partitions, η represents the solution dynamic viscosity, u represents the main direction velocity of the fluid, and v i represents the fluid flow rate of the i-th reconstruction image partition, k n represents the elastic coefficient of foam interaction, which can be found in the table by looking up the standard foam model. ij Indicates that δ ij represents the overlap between the i-th and j-th reconstruction image partitions, d j Represents the maximum main path of the j-th reconstruction image partition, r ij represents the centroid distance between the i-th and j-th reconstructed image partitions;

[0093] The surface tension is determined based on the interface length of the segmented region, the surface tension coefficient, and the normal direction outside the segmented region. Time advancement and dynamic evolution are performed on all segmented regions, which can be expressed as:

[0094] F su,i =r t ·L cu,i ·n i ;

[0095]

[0096] x i (t+Δt)=x i (t)+v i (t+Δt)·Δt;

[0097] Among them F su,i represents the surface tension of the i-th reconstruction image partition, r t Indicates the surface tension coefficient, which can be found in the table based on the concentration of the reagent / pH. cu,iIndicates the length of the i-th reconstruction partition interface, n i Indicates the normal direction outside the segmentation area, x i represents the center coordinate of the i-th reconstruction partition, and Δt represents the time step;

[0098] According to the position information and the maximum main diameter of each reconstructed image partition after advancement, grid mapping is performed to generate a simulated image, and the consistency error is verified with the high-resolution physical consistent reconstruction image, which is expressed as:

[0099]

[0100] Among them E dif Represents the average pixel difference value, M and N represent the number of pixels of height and width of the image respectively, Represents the pixel value at (x,y) of the high-resolution physically consistent reconstruction image, Represents the pixel value at (x,y) of the simulated image;

[0101] The sum of the historical mean and standard deviation is used as the difference threshold. If the average pixel difference value is less than the difference threshold, it means that the error is small.

[0102] By calculating the unit mass based on the maximum principal diameter of the partition and the foam density, the mass of each foam structure region in physical space is accurately estimated. The dynamic viscosity of the solution and the mainstream velocity are incorporated into the calculation of the fluid drag force, which can reasonably reflect the resistance of the fluid to the foam structure and enhance the realism of the fluid-foam interaction in the simulation. By calculating the collision force between the partitions, the elastic interaction within the foam group is simulated, effectively capturing the collision and squeezing phenomena between bubbles. This ensures that the foam movement and deformation during the simulation conform to the complex collective dynamic characteristics of the foam medium, improving the model's adaptability to the behavior of multi-foam systems. The surface tension is calculated by combining the interface length and normal direction of the partitions with the surface tension coefficient to accurately simulate the stretching and contraction forces of the foam membrane. Time advancement based on these forces is performed simultaneously on all partitions. By mapping the simulated partition position information and scale back to image space to generate simulated images, the simulated images are then compared with the high-resolution, physically consistent reconstruction images using quantitative methods such as the mean square error at the pixel level. This enables direct verification between the simulation results and the measured observations, enhancing the credibility and applicability of the simulation model.

[0103] S3, determine the combination of Reynolds number, foam volume fraction, foam coverage and surface tension coefficient as the main state vector, and construct the derivative equation of the change rate based on the main state vector, physical quantity vector and disturbance term.

[0104] Predict the rate of change, generate a contribution heat map, and adjust the proportional-integral (PI) control of the largest contributing term based on the consistency error verification results. After the adjustment, re-verify the consistency error.

[0105] Preferably, the Reynolds number, foam volume fraction, foam coverage and surface tension coefficient are determined as a main state vector, and a derivative equation of the change rate is constructed according to the main state vector, the physical quantity vector and the disturbance term to predict the change rate, including:

[0106] The ratio of the total area of ​​the segmented region to the total area of ​​the image is used as the foam volume fraction, and the Reynolds number is calculated based on the solution flow rate, solution density, and solution dynamic viscosity, which is expressed as:

[0107]

[0108] Among them, Re t represents the Reynolds number, p represents the solution density, Q t represents the solution flow rate, η represents the solution dynamic viscosity, which can be found in the table based on pH and temperature, and A represents the flow channel cross-sectional area;

[0109] The foam coverage is calculated by the ratio of the number of pixels in the foam area to the total number of pixels in the super-resolution image, and the foam volume fraction, Reynolds number, foam coverage and surface tension coefficient are combined into the main state vector;

[0110] The influence coefficient of the main process parameters on the state is determined according to the main state vector, physical quantity vector and temperature data at the corresponding time. At the same time, the image mechanics field coupling differential term is determined according to the gradient field of the reconstructed image partition and the main state vector. The derivative equation of the main state vector with respect to time is constructed in combination with the disturbance term to express the rate of change of the system state, which is expressed as:

[0111] F1=A·S t +B·P t +C·T t ;

[0112]

[0113] Among them, F1 represents the influence coefficient of the main process parameters on the state, F2 represents the differential term of the image mechanics field coupling, F3 represents the disturbance term, S t represents the main state vector, P t Represents a physical quantity vector, T t represents temperature data, A represents the influence coefficient of the main state vector on the change speed, B represents the influence coefficient of the physical quantity vector on the main state change, C represents the influence coefficient of temperature on the main state change, D represents the influence coefficient of the image gradient on the main state change, E represents the coupling influence coefficient between the main states, and G represents the influence coefficient of impurity disturbance on the main state change. represents the impurity concentration, represents the image gradient calculated from the high-resolution physically consistent reconstruction image using the Sobel operator method;

[0114] Using historical data and discrete difference approximation, the change rate of the main state is calculated, which is expressed as:

[0115]

[0116] Among them S t+Δt represents the main state vector that changes over time Δt+t, where Δt represents the time interval;

[0117] Construct data tables based on the main state vector, physical quantity vector, temperature, image gradient, and change rate, where each row represents data at a specific moment.

[0118] Use linear regression to fit and determine the values ​​of A, B, C, D, E, and G by the least squares method;

[0119] The fourth-order Runge-Kutta method RK4 is used for time advancement to update the main state vector, physical quantity vector, temperature, image gradient and predict the rate of change based on the derivative equation.

[0120] The foam volume fraction is calculated by using the ratio of the total area of ​​the segmented region to the area of ​​the entire image, and the Reynolds number is calculated by combining the flow rate, density and dynamic viscosity. This method is closely dependent on real-time images and process data to ensure that the main state quantity accurately reflects the physical characteristics of the process site, while avoiding dependence on complex manual measurements, improving the timeliness and automation level of the measurement. By combining the foam volume fraction, Reynolds number, coverage and surface tension coefficient as the main state vector, the model can more comprehensively capture the multi-dimensional information of the interaction between foam and fluid. Combining multi-source physical and image information such as main process parameters, temperature and image gradient, a change rate derivative equation is constructed, which not only reflects the dynamic influence of the main state on itself In addition, multi-scale coupling and external disturbances are introduced, so that the model can respond to real-time changes in environmental and process conditions and fully reflect the complex nonlinear dynamic characteristics of the system. By applying historical data and adopting the discrete difference approximation method to accurately calculate the main state change rate, a solid numerical foundation is provided for the regression fitting of the coefficient matrix, ensuring that the parameters obtained by the least squares method reflect the actual process history and physical laws, improving the fitting quality and generalization ability of the model, and applying the fourth-order Runge-Kutta method to time advance the derivative equation, which not only ensures the high accuracy and stability of numerical calculations, but also supports continuous dynamic simulation of the main state vector and its influencing factors, so as to predict the evolution trend of future states.

[0121] Preferably, a contribution heat map is generated, and the proportional integral PI control adjustment is performed on the largest contribution item in combination with the consistency error verification result. After the adjustment, the consistency error verification is performed again, including:

[0122] According to the change rate calculation of the derivative equation, the influence of the main process parameters on the state, the image mechanics field coupling differential term and the disturbance term on the main state change rate are normalized and calculated as the contribution;

[0123] According to the main state vector and contribution value, a heat map data table is organized and a heat map is drawn using a drawing tool. The sum of the historical mean and the standard deviation is used as the contribution threshold for each contribution. If the contribution of the data item is greater than or equal to the contribution threshold, it is determined that the data item needs to be adjusted, and proportional integral PI control is used for adjustment. After adjustment, the consistency error verification is performed again.

[0124] By normalizing the rates of change of different terms in the derivative equation and quantifying their contributions to the main state changes, we can effectively achieve a detailed division and dynamic monitoring of the internal mechanisms of the process system. This contribution analysis can clearly distinguish the actual impact of the main process parameters, image mechanics field coupling, and disturbance terms on the system state changes. The contribution is combined with the main state vector and organized into a structured data table. Presented through a heat map, it enhances the visualization of complex multidimensional data, enabling operators and managers to intuitively understand the relative importance of each influencing factor and its multi-time evolution trend.

[0125] S4, record operation data to build a historical abnormal case library;

[0126] Furthermore, the operation data is recorded to build a historical abnormal case library, including:

[0127] If the average pixel difference value is greater than or equal to the difference threshold, it means that the error is large, and the main state vector, physical quantity vector, and temperature data are recorded as abnormal data to build a historical abnormal case library.

[0128] By setting the comparison between the average pixel difference value and the difference threshold, it helps to effectively identify abnormal data in cases with large errors, ensuring that abnormal phenomena can be detected and recorded in a timely manner. Recording the main state vector, physical quantity vector and temperature data not only enriches the data dimension of the abnormal case library, but also provides a multi-angle reference basis for subsequent analysis, improving the accuracy and reliability of abnormal diagnosis. Building a historical abnormal case library lays a solid foundation for model training and abnormal pattern recognition, helps to optimize the system's anomaly detection algorithm, and thus improves the overall stability and security of the system.

[0129] This embodiment also provides a computer control system for a flotation production process, including:

[0130] Data processing module, which obtains multimodal data of the flotation cell through sensors and acquisition equipment;

[0131] Image reconstruction module, builds an image super-resolution model, generates a high-resolution physically consistent reconstruction image, and calculates the maximum main diameter of the segmented area and the overall distribution mean;

[0132] The evolution simulation module promotes the dynamic evolution of all segmented areas in the time dimension, generates simulated images of foam behavior, and uses the simulated images to verify consistency errors;

[0133] The rate prediction module calculates fluid dynamics related parameters, generates the main state vector, constructs the derivative equation of the change rate, and predicts the change rate of foam and material flow;

[0134] The contribution monitoring module calculates the impact of the main process parameters on the system, generates a heat map of the impact contribution, and identifies anomalies based on the results of consistency error verification.

[0135] This embodiment also provides a computer device suitable for the computer control method of the flotation production process, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the computer control method of the flotation production process proposed in the above embodiment.

[0136] The computer device may be a terminal, comprising a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner may be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a button, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse.

[0137] This embodiment further provides a storage medium having a computer program stored thereon. When the program is executed by a processor, the computer control method for a flotation production process as proposed in the above embodiment is implemented. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0138] In summary, the present invention enhances and corrects the neighborhood pixels by combining calibrated defect correction parameters, which not only improves the image detail level and signal-to-noise ratio, but also corrects the systematic errors caused by the equipment itself, so that the super-resolution input image quality is higher, the spatial detail information of the image is richer and the physical consistency is stronger. The watershed algorithm is used to segment the foam area of ​​the enhanced image, which can effectively identify independent foam structures that are not connected to each other in the image. The alignment of timestamps and physical quantity vectors enables the physical parameters to be accurately mapped to specific foam areas, ensuring the authenticity and supervision effect of the training samples. By calculating the collision force between the segmented areas and simulating the elastic interaction within the foam group, the collision and extrusion phenomena between foams are effectively captured, so that the foam movement and deformation during the simulation process conform to the complex collective dynamic characteristics of the foam medium. By constructing the rate of change derivative equation, it not only reflects the dynamic influence of the main state on itself, but also introduces multi-scale coupling and external disturbances, so that the model can respond to real-time changes in environmental and process conditions, and fully reflect the complex nonlinear dynamic characteristics of the system.

[0139] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A computer control method for a flotation production process, characterized in that: include: The system collects multimodal data in the flotation tank, including graphic frame data and physical quantity vectors, calculates the foam reflectivity to determine the light source brightness and color temperature adjustment coefficient, performs enhancement correction, and constructs an image super-resolution model to output a high-resolution physically consistent reconstruction image. Calculate the unit mass of the reconstructed image and the collision force between partitions. Perform time advancement and dynamic evolution on all partitioned regions and verify the consistency error. Determine the Reynolds number, foam volume fraction, foam coverage, and surface tension coefficient combination as the main state vector. Construct a derivative equation for the rate of change based on the main state vector, physical quantity vector, and disturbance term. Predict the rate of change and generate a contribution heat map. Combined with the consistency error verification results, perform proportional-integral (PI) control adjustments on the term with the largest contribution. After the adjustments, re-verify the consistency error. Record operation data to build a historical abnormal case library.

2. The computer control method for flotation production process according to claim 1, characterized in that: The output high-resolution physical consistent reconstruction map includes: Use a light meter to read the light intensity on the foam surface, and use the collected graphic frame data to measure the foam grayscale at the center of the foam area, and determine the grayscale mean of the actual grayscale area of ​​the foam to determine the foam reflectivity; Based on the standard light intensity reference value and standard color temperature reference value of the foam area determined by historical experience, the multi-spectral lighting compensation coefficient is calculated, including the adjustment coefficients of brightness and color temperature; Adjustment coefficients based on brightness and color temperature are input and adjusted through the LED light power control to complete dimming and color adjustment, and light correction graphic frame data is re-collected to perform enhanced correction based on field pixels; The watershed algorithm is used to segment the enhanced graphic frame data to obtain different independent foam partitions at the same time; The enhanced image frame data is aligned with the normalized physical quantity vectors for timestamp data, which is then fed into the image super-resolution model. The model is then trained on historical enhanced images and high-resolution annotated images, where each sample is labeled with a physical parameter. Output a high-resolution physically consistent reconstruction image based on the trained image super-resolution model; The watershed algorithm is used again to segment the high-resolution physically consistent reconstruction image. Within the i-th segmented subdomain after re-segmentation, the two point pairs with the largest Euclidean distance are found, and the pixel Euclidean distance of the point pair coordinates is determined as the maximum main path of the reconstructed image partition. The overall distribution mean is calculated based on the major axis lengths of the reconstructed image partitions and the total number of segmented regions.

3. The computer control method for flotation production process according to claim 2, characterized in that: The calculation reconstructs the unit quality of the image, calculates the collision force between partitions, performs time advancement and dynamic evolution on all partitioned regions, and verifies the consistency error, including: According to the normalized physical quantity vector, the simulation environment is built, the simulation space size is determined according to the high-resolution physical consistent reconstruction map, and the unit mass is preliminarily calculated according to the standard value of foam density; Synchronously calculate the fluid drag force by combining the solution dynamic viscosity and the main direction velocity of the fluid, and calculate the collision force between partitions; Determine the surface tension based on the interface length of the segmented region, the surface tension coefficient, and the normal direction outside the segmented region, and perform time advancement and dynamic evolution on all segmented regions. Grid mapping is performed based on the position information and maximum main diameter of each reconstructed image partition after advancement, and a simulation image is generated. The consistency error is verified with the high-resolution physical consistent reconstruction image.

4. The computer control method for flotation production process according to claim 3, characterized in that: The method comprises: determining the Reynolds number, foam volume fraction, foam coverage and surface tension coefficient as the main state vector, constructing a derivative equation of the change rate according to the main state vector, the physical quantity vector and the disturbance term, and predicting the change rate, including: The ratio of the total area of ​​the segmented region to the total area of ​​the image is used as the foam volume fraction, and the Reynolds number is calculated based on the solution flow rate, solution density and solution dynamic viscosity; The foam coverage is calculated by the ratio of the number of pixels in the foam area to the total number of pixels in the super-resolution image, and the foam volume fraction, Reynolds number, foam coverage and surface tension coefficient are combined into the main state vector; The influence coefficients of the main process parameters on the state are determined based on the main state vector, physical quantity vector, and temperature data at the corresponding time. At the same time, the image mechanics field coupling differential terms are determined based on the gradient field of the reconstructed image partition and the main state vector. Combined with the disturbance term, the derivative equation of the main state vector with respect to time is constructed to represent the rate of change of the system state. Use historical data and discrete difference approximation to calculate the rate of change of the main state; Construct data tables based on the main state vector, physical quantity vector, temperature, image gradient, and change rate, where each row represents data at a specific moment. The fourth-order Runge-Kutta method RK4 is used for time advancement to update the main state vector, physical quantity vector, temperature, image gradient and predict the rate of change based on the derivative equation.

5. The computer control method for flotation production process according to claim 4, characterized in that: The contribution heat map is generated, and the proportional integral PI control adjustment is performed on the largest contribution item in combination with the consistency error verification result. After the adjustment, the consistency error verification is performed again, including: According to the change rate calculation of the derivative equation, the influence of the main process parameters on the state, the image mechanics field coupling differential term and the disturbance term on the main state change rate are normalized and calculated as the contribution; According to the main state vector and contribution value, a heat map data table is organized and a heat map is drawn using a drawing tool. The sum of the historical mean and the standard deviation is used as the contribution threshold for each contribution. If the contribution of the data item is greater than or equal to the contribution threshold, it is determined that the data item needs to be adjusted, and proportional integral PI control is used for adjustment. After adjustment, the consistency error verification is performed again.

6. The computer control method for flotation production process according to claim 5, characterized in that: The multimodal data collected in the flotation cell includes: Multi-angle, high-frame-rate industrial CCD+spectral enhancement cameras are deployed simultaneously at key locations in the flotation tank to acquire multimodal data on the 3D morphology and texture of the foam surface, as well as process data such as flow rate, pH, conductivity, bubble size distribution, and reagent concentration, as well as graphic frame data. The collected real-time flow rate of flotation cell feed, pH value of flotation cell slurry, slurry conductivity, average diameter of foam in the foam zone, and online concentration of reagent are preprocessed and timestamped to form a physical quantity vector.

7. The computer control method for flotation production process according to claim 6, characterized in that: The recorded operation data is used to build a historical abnormal case library, including: If the average pixel difference value is greater than or equal to the difference threshold, it means that the error is large, and the main state vector, physical quantity vector, and temperature data are recorded as abnormal data to build a historical abnormal case library.

8. A computer control system for a flotation production process, based on the computer control method for a flotation production process according to any one of claims 1 to 7, characterized in that: include, Data processing module, which obtains multimodal data of the flotation cell through sensors and acquisition equipment; Image reconstruction module, builds an image super-resolution model, generates a high-resolution physically consistent reconstruction image, and calculates the maximum main diameter of the segmented area and the overall distribution mean; The evolution simulation module promotes the dynamic evolution of all segmented areas in the time dimension, generates simulated images of foam behavior, and uses the simulated images to verify consistency errors; The rate prediction module calculates fluid dynamics related parameters, generates the main state vector, constructs the derivative equation of the change rate, and predicts the change rate of foam and material flow; The contribution monitoring module calculates the impact of the main process parameters on the system, generates a heat map of the impact contribution, and identifies anomalies based on the results of consistency error verification.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the computer control method for a flotation production process according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the computer control method for a flotation production process according to any one of claims 1 to 7 are implemented.

Images