A thickener feedwell dilution flocculation intelligent control system and device
By introducing a level difference-enhanced baffle and a self-dilution water channel into the thickener feed well, combined with an intelligent control system, the problem of unreasonable dosage and location during the dilution and flocculation process in traditional thickener feed wells has been solved. This has enabled precise flocculant addition, improved overflow water quality stability, and reduced operating costs.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- HUAIBEI MINE MASCH MFG CO LTD
- Filing Date
- 2026-02-26
- Publication Date
- 2026-07-10
AI Technical Summary
In the traditional thickener feeding well dilution and flocculation process, it is difficult to accurately control the dosage and the addition position is unreasonable, resulting in substandard dilution concentration, high operating costs, and poor mixing effect under the influence of multiple factors.
A thickener feed well dilution and flocculation intelligent control system and device is adopted. Through the design of liquid level difference enhanced baffle and self-dilution water channel, combined with data acquisition, model prediction and feedback adjustment, an intelligent flocculant dosing strategy is realized to ensure flocculation effect.
It improves the stability of overflow water quality, reduces flocculant dosage, reduces operating costs, and can effectively cope with disturbances from multiple factors, thereby improving the stability and reliability of the dilution and flocculation process in the thickener feed well.
Smart Images
Figure CN121846738B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of flocculation technology for thickener feed wells, specifically to an intelligent control system and device for dilution and flocculation in thickener feed wells. Background Technology
[0002] In industries such as mining and metallurgy, thickeners are core equipment for solid-liquid separation. Their working efficiency and separation effect directly affect production continuity and product quality, and the dilution and flocculation effect of the feed well is the key to determining the overall performance of the thickener. The location of flocculant addition has a significant impact on the flocculation effect of slurry: too far away will cause floc disintegration, while too close will result in insufficient mixing and easy waste of reagents. Therefore, under varying operating conditions, how to intelligently and accurately adjust the operating parameters of the feed well to achieve stable overflow water quality and increased underflow concentration has become an urgent technical problem to be solved.
[0003] Traditional thickener feed wells currently face multiple challenges in the dilution and flocculation process: Structurally, single-well structures have low momentum dissipation intensity, making it difficult to effectively dissipate the kinetic energy of the slurry. Furthermore, the circulation effect created by tangential feeding raises the liquid level inside the well, reducing the self-dilution level difference between the inside and outside of the feed well, resulting in insufficient dilution water. In terms of dilution and flocculation control, due to the dynamic changes in parameters such as the composition, particle size, and flow rate of the incoming slurry, as well as its large time-delay response characteristics, and the disturbances caused by factors such as turbulence intensity, temperature, pH value, and particle characteristics, existing systems struggle to accurately control the flocculant dosage. Moreover, the fixed dosage location fails to fully consider the fluid dynamics characteristics inside the well, making it impossible to adapt to different operating conditions and resulting in poor mixing. Summary of the Invention
[0004] The purpose of this invention is to provide an intelligent control system and device for dilution and flocculation in the feed well of a thickener, in order to solve the problems of difficulty in accurately controlling the dosage, unreasonable addition location, failure to achieve optimal dilution concentration, high operating costs, and disturbances from multiple factors during the flocculant addition process in the feed well of a thickener. The invention aims to achieve an intelligent flocculant addition strategy, ensure flocculation and thickening effect, and reduce operating costs.
[0005] To achieve the above objectives, the present invention provides the following technical solution: a thickener feed well dilution and flocculation intelligent control device, comprising a slurry inlet pipe, a diversion speed regulating tank, and a main feed well. The slurry inlet pipe enters the diversion speed regulating tank tangentially. Multiple diversion branch pipes are evenly arranged at the bottom of the diversion speed regulating tank and extend into the main feed well. A liquid level difference strengthening baffle is set at a height close to the liquid level in the upper part of the upper feed well, and multiple self-dilution water channels are opened on the circumferential side wall. A flocculant solution adding ring pipe is set in an annular zone between the outer side of the diversion speed regulating tank and the inner side of the main feed well. The flocculant solution adding ring pipe is located in the middle position above the liquid level difference strengthening baffle. Multiple flocculant solution nozzles are evenly arranged in the circumferential direction at the lower part of the flocculant solution adding ring pipe.
[0006] The outside of the self-dilution water channel is connected to the overflow water of the thickener, and a dilution water volume regulating gate and a drive connecting rod are installed at the outer end.
[0007] The inner wall of the main feed well is equipped with an annular baffle, and the discharge ports at the ends of the multiple branch pipes are located between the liquid level difference enhanced baffle and the annular baffle.
[0008] The liquid level difference enhanced turbulence plate is composed of multiple inclined plates, which are evenly arranged in the circumferential direction inside the feed well and form an angle of 10-25° with the tangential direction of the slurry circulation direction. At the same time, it is ensured that its top is located about 10-30mm below the self-dilution water channel.
[0009] The liquid level difference enhanced turbulence plate consists of 6-12 inclined plates, and the vertical projection width is 1 / 8-1 / 5 of the feed well diameter.
[0010] The present invention also provides an intelligent control system for optimizing the dilution and flocculation effect of a thickener feed well, comprising: a data acquisition layer, a model prediction layer, an operation execution layer, and a feedback adjustment layer.
[0011] Furthermore, the data acquisition layer includes a flow meter and a concentration meter installed on the slurry inlet pipe, and a level gauge installed above the flow diversion speed regulating tank. The operation execution layer includes a flocculant solution addition loop connected to the front end of a controllable flow flocculant addition pump, multiple flocculant solution nozzles connected to a control system and each nozzle can be controlled individually, and a drive connecting rod above each dilution water volume regulating gate connected to a power drive device. The feedback adjustment layer includes a turbidity meter and a concentration meter installed on the overflow and underflow of the thickener, respectively.
[0012] Furthermore, the intelligent control strategy for optimizing the dilution and flocculation effect of the thickener feed well of the device includes the following four steps: data acquisition, model prediction, operation execution, and feedback adjustment:
[0013] S1: Data Acquisition: Using a data acquisition module, including instruments such as flow meters, concentration meters, turbidity meters, and level gauges, historical data of auxiliary parameters such as feed slurry concentration, flow rate, and level are collected and defined as X. i The corresponding data for overflow turbidity and underflow concentration based on collaborative feedback are defined as Z. i The data is cleaned, missing and outlier values are removed, and standardization methods are used to preprocess the data so that all features have the same scale.
[0014] S2: Model Prediction: Using machine learning and other methods, an optimal mathematical model that matches the expected results is pre-established based on experimental data. The thresholds of each parameter in the prediction model are then adjusted. The result calculated according to the prediction parameters is defined as Y. i The self-dilution water volume, flocculant solution addition amount, and addition location were set respectively.
[0015] S3: Operation Execution: Based on the prediction results, the control and execution module sends control signals to relevant actuators to adjust the dilution water volume, flocculant addition location, and addition amount. Specifically, it sets the appropriate self-dilution water volume by controlling the opening and closing degree of the dilution water regulating gate; it precisely determines the flocculant addition amount through actuators such as the flocculant addition pump; and it selects the appropriate addition location by controlling the flocculant solution nozzle switches at different positions.
[0016] S4: Feedback Adjustment: The monitoring and feedback module monitors the actual overflow turbidity and underflow concentration in real time, compares them with the predicted values, and if the deviation exceeds the set threshold, collects data again, updates and trains the model, and optimizes the control strategy.
[0017] Slight deviation is defined as a difference of ≤5% between the actual and predicted values, in which case only the model's penalty factor C is adjusted; severe deviation is defined as a difference of >10%, in which case at least 100 sets of real-time data are collected again for model training, and the prediction accuracy is improved by ≥15% after training. This solves the deviation caused by large time delays and multiple perturbations.
[0018] Furthermore, the principle of the intelligent control strategy for optimizing the dilution and flocculation effect of the thickener feed well of the device is described as follows:
[0019] The real-time data acquired by the data acquisition layer is defined as follows: level gauge data is X1, feed concentration meter data and flow meter data are X2 and X3, respectively. The real-time data acquired by the feedback adjustment layer is used as the control target, and the overflow turbidity meter is defined as Z1, and the underflow concentration meter as Z2. It should be noted that the underflow concentration can be calculated using the pressure value obtained from the underflow pressure meter or used as the control target; the underflow concentration meter is only used as an example here. Simultaneously, the setting parameters of the operation execution layer are defined: flocculant addition amount and addition location are Y1 and Y2, respectively, and the self-dilution gate opening degree is Y3.
[0020] Based on the analysis of influencing factors, the interrelationships of the various defined parameters are established, namely, the functional relationship that satisfies the following equation:
[0021]
[0022] In this embodiment, the data analysis and prediction module uses grid search, random search, or Bayesian optimization methods to fine-tune the penalty factor C and kernel coefficient γ of the support vector machine regression model.
[0023] In the feedback adjustment step, different deviation threshold levels are set, and different feedback and optimization measures are taken according to the severity of the deviation. For example, when there is a slight deviation, only the model parameters are fine-tuned, and when there is a severe deviation, a large amount of data is collected again to fully train the model.
[0024] In the above technical solution, the intelligent control system and device for dilution and flocculation in the feed well of a thickener provided by the present invention has the following beneficial effects:
[0025] Compared to traditional methods, the fluctuation range of overflow water clarity is reduced, effectively reducing overflow water turbidity and improving the stability of overflow water quality. Through intelligent control model for precise matching of dosage and independent controllable nozzles for dosage location, combined with dynamic adjustment of self-dilution water volume, the amount of flocculant added is reduced while meeting the underflow and overflow indicators. It can also effectively cope with dynamic changes in incoming slurry parameters and multi-factor disturbances, solve time lag problems, and improve the stability and reliability of the dilution and flocculation process in the thickener feed well. Attached Figure Description
[0026] To more clearly and accurately explain the specific embodiments of the present invention, the accompanying drawings used in the description of the specific embodiments and the prior art are briefly introduced below. It is obvious that the drawings mentioned in the following description are those associated with certain specific embodiments of the present invention. In the drawings, the elements or parts are not necessarily drawn to scale. For those skilled in the art, other related drawings can be obtained based on these drawings without any creative intellectual effort, to assist in a comprehensive understanding and application of the technical solutions of the present invention.
[0027] Figure 1 A schematic diagram of the overall structure of the thickener feed well is shown;
[0028] Figure 2 A schematic diagram of the internal structure of the thickener feed well is shown;
[0029] Figure 3 A schematic diagram of the longitudinal cross-sectional structure of the thickener feed well is shown;
[0030] Figure 4 A schematic diagram of the self-dilution channel structure of the thickener feed well is shown;
[0031] Figure 5 A schematic diagram of the loop structure for adding flocculant solution is shown.
[0032] Figure 6 A schematic diagram of the feed well dilution and flocculation optimization strategy system is shown;
[0033] Figure 7 The diagram shows the logic principle of the feed well dilution and flocculation optimization strategy;
[0034] Figure 8 A comparison chart of the debugging and operation of the embodiment is shown.
[0035] Reference numerals: 1. Slurry inlet pipe; 2. Diversion speed regulating tank; 21. Diversion branch pipe; 3. Thickener main feed well; 31. Liquid level difference reinforced baffle plate; 32. Annular baffle plate; 4. Self-dilution water channel; 41. Dilution water volume regulating gate; 42. Drive connecting rod; 5. Flocculant solution addition ring pipe; 51. Flocculant solution nozzle. Detailed Implementation
[0036] To enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings.
[0037] Please see Figure 1-8 This invention provides an intelligent control device for dilution and flocculation in a thickener feed well, comprising a slurry inlet pipe 1, a diversion speed regulating tank 2, and a main feed well 3. The slurry inlet pipe 1 enters the diversion speed regulating tank 2 tangentially. Multiple diversion branch pipes 21 are evenly arranged at the bottom of the diversion speed regulating tank 2 and extend into the main feed well 3. A liquid level difference strengthening baffle 31 is set at a height close to the liquid level at the upper part of the upper feed well 3, and multiple self-dilution water channels 4 are opened on the circumferential side wall. A flocculant solution adding ring pipe 5 is set in an annular zone between the outer side of the diversion speed regulating tank 2 and the inner side of the main feed well 3. The flocculant solution adding ring pipe 5 is located in the middle position above the liquid level difference strengthening baffle 31. Multiple flocculant solution nozzles 51 are evenly arranged in the circumferential direction at the lower part of the flocculant solution adding ring pipe 5.
[0038] In another embodiment of the present invention, the spray angle of the flocculant solution nozzle 51 is 45-60°, facing the central axis of the main feed well 3. Multiple flocculant solution nozzles 51 can switch between multiple sets of different position combinations according to the real-time data of feed flow rate and concentration. In this way, the mixing uniformity of flocculant and slurry is ≥92%, which is 40% higher than that of fixed nozzle structure.
[0039] In another embodiment of the present invention, the outside of the self-dilution water channel 4 is connected to the overflow water of the thickener, and a dilution water volume regulating gate 41 and a drive connecting rod 42 are provided at the outer end;
[0040] The inner wall of the main feed well 3 is provided with an annular baffle 32, and the discharge ports at the ends of the multiple branch pipes 21 are located between the liquid level difference reinforced baffle 31 and the annular baffle 32.
[0041] The discharge port at the end of the branch pipe 21 forms an angle of 15-30° with the central axis of the main feed well 3. After the slurry is diverted, it is adapted to the flow field formed by the annular baffle 32. When the slurry enters the main feed well 3, the flow velocity fluctuation is ≤5% and the distribution uniformity is ≥90%.
[0042] In another embodiment of the present invention, the liquid level difference enhanced turbulence plate 31 is composed of multiple inclined plates, which are evenly arranged in the circumferential direction inside the feed well and form an angle of 10-25° with the tangential direction of the slurry circulation direction. At the same time, it is ensured that its top end is located about 10-30mm below the self-dilution water channel 4.
[0043] The liquid level difference enhanced turbulence plate 31 consists of 6-12 inclined plates, and its vertical projection width is 1 / 8-1 / 5 of the diameter of the feed well.
[0044] By enhancing the setting of the baffle plate 31 through the liquid level difference, a liquid level difference is generated on the inner and outer sides of the feed well 3. The included angle of the baffle plate in the prior art is mostly 30-45°, and the positional relationship with the self-dilution water channel is not limited. Compared with the prior art, the present invention has a 30% improvement in effect.
[0045] In summary, the core breakthrough of this invention mainly targets the internal flow field characteristics of the thickener feed well: First, through computational fluid dynamics simulation, it was found that when the traditional feed well adopts tangential feeding, the slurry forms a strong circulation along the well wall under the action of centrifugal force, which leads to an overall increase in the liquid level inside the well and a small liquid level difference with the external overflow area. The self-dilution water cannot fully penetrate due to insufficient pressure difference, resulting in uneven dilution. Further analysis of the flow field vector distribution shows that the kinetic energy of the circulation is concentrated on the outside of the well wall, while the fluid velocity inside is low and the momentum dissipation is weak. This not only affects the flocculation reaction but also exacerbates the vicious cycle of insufficient liquid level difference.
[0046] To address this issue, the initial plan was to use the conventional approach of increasing the diameter of the dilution water channel. However, simulations revealed that while this would increase the flow rate, it would disrupt the stability of the slurry flow pattern, leading to floc disintegration. Therefore, based on the core logic of flow field control, we needed to both dissipate the circulating kinetic energy and guide the fluid to form an effective pressure difference. This led us to consider setting up a turbulence structure in the concentrated circulation region.
[0047] The level difference enhancement baffle 31 needs to be at a certain angle to the tangential direction of the circulation. This is necessary to cut the circulation and dissipate its kinetic energy to lower the liquid level in the well, while avoiding excessive disturbance to the slurry to prevent floc destruction. Through multiple flow field simulation iterations, it was found that an angle of 10-25° is the optimal range: when it is less than 10°, the cutting effect on the circulation is weak and the level difference is not sufficiently increased; when it is greater than 25°, local turbulence will occur and the floc dissociation rate will exceed 10%. Therefore, this angle range was determined.
[0048] The water flow in the self-dilution water channel 4 needs to enter the well with the help of the liquid level difference. If the top of the liquid level difference strengthening baffle 31 is higher than the channel outlet, it will block the water flow. If it is too much lower than the channel outlet, it cannot effectively act on the circulation core area. By simulating the water flow infiltration efficiency and liquid level difference changes at different positions, it was determined that the top is located 10-30mm below the self-dilution water channel. This ensures the smooth flow of dilution water and guides the water flow to form an orderly mixture with the slurry through the baffle, thereby increasing the liquid level difference and completely solving the problem of insufficient dilution.
[0049] If the vertical projection width of the liquid level difference enhanced baffle 31 is too large, it will occupy the slurry reaction space; if it is too small, it will not be able to cover the circulation area. Based on the matching analysis of the feed well diameter and the circulation influence range, it is determined that its width does not exceed 1 / 5 of the well diameter, so as to achieve the flow field control target without interfering with the overall flow of the slurry.
[0050] An intelligent control system for optimizing the dilution and flocculation effect of a thickener feed well includes: a data acquisition layer, a model prediction layer, an operation execution layer, and a feedback adjustment layer.
[0051] The data acquisition layer includes a flow meter and a concentration meter installed on the slurry inlet pipe, and a level gauge installed above the flow diversion speed regulating tank. The operation execution layer includes a flocculant solution addition loop connected to the front end of a controllable flow flocculant addition pump, multiple flocculant solution nozzles connected to a control system and each nozzle can be controlled individually, and a drive connecting rod above each dilution water volume regulating gate connected to a power drive device. The feedback adjustment layer includes a turbidity meter and a concentration meter installed on the overflow and underflow of the thickener, respectively.
[0052] It includes the following four steps: data collection, model prediction, operation execution, and feedback adjustment:
[0053] S1: Data Acquisition: Historical data of feed slurry concentration, flow rate and liquid level auxiliary parameters are collected using flow meters, concentration meters, turbidity meters and liquid level meters in the data acquisition module and defined as Xi. The corresponding data of overflow turbidity and underflow concentration are defined as Zi. The data is cleaned, missing values and outliers are processed, and the data is preprocessed using a standardization method to make each feature have the same scale.
[0054] S2: Model Prediction: Using machine learning methods, the optimal mathematical model that matches the results is pre-established based on experimental data, and the thresholds of each parameter of the prediction model are adjusted. The calculation results of the prediction parameters are defined as Yi, and the self-dilution water volume, flocculant solution addition amount and addition location are set respectively.
[0055] S3: Operation Execution: Based on the prediction results, the control execution module sends control signals to the relevant execution mechanisms to adjust the dilution water volume, the location and amount of flocculant added;
[0056] That is, the appropriate self-dilution water volume is set by controlling the opening and closing degree of the gate to adjust the dilution water volume;
[0057] The amount of flocculant added is precisely controlled by actuators such as flocculant addition pumps, and the appropriate addition position is selected by controlling the flocculant solution nozzles at different locations.
[0058] S4: Feedback Adjustment: The monitoring and feedback module monitors the actual overflow turbidity and underflow concentration in real time, compares them with the predicted values, and if the deviation exceeds the set threshold, collects data again, updates and trains the model, and optimizes the control strategy.
[0059] The principle of the intelligent control strategy is described as follows:
[0060] The real-time data acquired by the data acquisition layer is defined as follows: level gauge data is X1, feed concentration meter data and flow meter data are X2 and X3, respectively. The real-time data acquired by the feedback adjustment layer is used as the control target, and the overflow turbidity meter is defined as Z1, and the underflow concentration meter as Z2. It should be noted that the underflow concentration can be calculated using the pressure value obtained from the underflow pressure meter or used as the control target; the underflow concentration meter is only one example here. Simultaneously, the setting parameters of the operation execution layer are defined: flocculant addition amount and addition location are Y1 and Y2, respectively, and the self-dilution gate opening degree is Y3.
[0061] Based on the analysis of influencing factors, the interrelationships of the various defined parameters are established, namely, the functional relationship that satisfies the following equation:
[0062]
[0063] In another embodiment of the present invention, the data analysis and prediction module uses one or more of the following methods—grid search, random search, and Bayesian optimization—to fine-tune the penalty factor C and kernel coefficient γ in the support vector machine regression model parameters.
[0064] In the feedback adjustment step, different deviation threshold levels are set, and different feedback and optimization measures are taken according to the severity of the deviation. For example, when there is a slight deviation, only the model parameters are fine-tuned, and when there is a severe deviation, a large amount of data is collected again to fully train the model.
[0065] Slight deviation is defined as a difference of ≤5% between the actual and predicted values. Only the penalty factor C of the model is adjusted to quickly respond to short-term operating condition fluctuations. Severe deviation is defined as a difference of >10%. At least 100 sets of real-time data are collected again for model training, and the parameter mapping relationship is updated to ensure long-term control accuracy. The prediction accuracy after training is improved by ≥15%. The feedback mechanism can solve the problems of large time delays and multiple disturbances.
[0066] In embodiments of the present invention, a radar level gauge is installed above the distribution tank of the feed well, a flow meter and a concentration meter are installed on the feed pipe, a low-range turbidity meter is installed at the outlet of the thickener overflow weir, and a concentration meter is installed on the underflow pipe of the thickener tank. The above instruments are calibrated and adjusted. The hardware and software environment for an intelligent control system for optimizing dilution and flocculation effects is established, and all devices and systems are connected and overall debugging is performed to ensure accurate data transmission and stable system operation.
[0067] Data Acquisition and Model Training: After the system is running, the online monitoring instrument continuously collects various data, including real-time data such as feed flow rate, feed concentration, and liquid level. The collected historical data is organized and normalized, and then input into the support vector machine regression model for training iterations. By finding the optimal hyperplane, the overflow turbidity and underflow concentration of the thickener are predicted. After multiple training and optimizations, the model achieves high prediction accuracy. In this embodiment, the model boundary conditions are set as follows: overflow turbidity fZ1 ≤ 200 PPM, and thickener underflow concentration fZ2 ≥ 60%.
[0068] The following mainly involves the actions of the implementing agency, feedback and adjustments during the execution process, and the coordination between the execution operation and other links. The execution operation part is described in detail below:
[0069] Based on decision commands, the system precisely adjusts the flocculant dosage by controlling the flocculant addition equipment. If the decision is to increase the flocculant dosage, the metering pump will increase the pumping frequency or stroke to inject more flocculant solution into the feed well through the addition loop. Conversely, it will decrease the pumping frequency or stroke to reduce the dosage. For example, when the system detects an increase in slurry concentration and poor flocculation, the decision module issues a command for the metering pump to increase the flocculant dosage by 20% to enhance the flocculation effect.
[0070] Precise control of dilution water flow is achieved by changing the opening of an electric regulating valve. If an increase in dilution water flow is needed, the opening of the electric regulating valve is increased, and vice versa. For example, when the concentration of the feed slurry exceeds the set range, the decision system issues a command to increase the opening of the electric regulating valve by 30%, allowing more self-dilution water to enter the feed well, reducing the slurry concentration and optimizing the flocculation environment.
[0071] During operation, sensors continuously monitor key parameters within the feed well, such as slurry concentration, distribution tank level, and overflow turbidity. This real-time data is fed back to the control system to assess the impact of the operation on the dilution and flocculation effect.
[0072] For example, if the overflow turbidity is monitored by a turbidity meter and the overflow turbidity does not decrease to the expected range after increasing the amount of flocculant added, it indicates that the effect of the operation has not reached the ideal state.
[0073] Based on feedback data, the control system analyzes deviations in the executed operations and makes dynamic adjustments. If the slurry concentration still does not reach the set value after adjusting the dilution water flow rate, the system will further fine-tune the opening of the electric regulating valve.
[0074] If the flocculation effect does not improve significantly or even deviates significantly after adjusting the amount and location of flocculant addition, the system will comprehensively consider the iterative prediction model, collect a large amount of data to fully train and optimize the model, recalculate and adjust the flocculant addition strategy to ensure that the control operation is precise and effective.
[0075] The results of the operation provide new data samples for data collection. These data are then analyzed to evaluate the effectiveness of the control strategy. If the flocculant dosage and dilution water flow rate are frequently adjusted during a certain period, the analysis can identify potential problems or optimization directions through in-depth data mining, providing a more reliable basis for subsequent decision-making and achieving continuous optimization of the control strategy.
[0076] The decision-making process adjusts its logic and parameters promptly based on feedback from the executed operations. If certain decision rules prove ineffective under specific conditions after execution, the process will re-evaluate these rules, adjust decision thresholds, or add new decision conditions to better align decisions with actual production needs, ensuring the efficient and stable operation of the entire intelligent control system.
[0077] During normal system operation, the liquid level in the diversion speed regulating tank is monitored in real time, and the dilution water volume is adjusted accordingly. The amount and location of flocculant addition are intelligently controlled. Compared with traditional manual dosing strategies, the precise dosing control system of this invention significantly reduces the fluctuation in overflow water clarity and the amount of flocculant added is significantly reduced. While ensuring the concentration of the thickener underflow and the quality of the overflow water, the expected technical effects and economic benefits are achieved.
[0078] Compared with existing technologies, the core drawback of traditional control schemes is that they treat parameters such as feed flow rate, concentration, and flocculant dosage as independent variables, ignoring their inter-coupling characteristics. For example, when the feed concentration increases, not only is it necessary to increase the flocculant dosage, but also to adjust the dilution water volume and addition position simultaneously. Only through the synergy of these three factors can the flocculation effect be guaranteed. Furthermore, the large time lag in the thickener flocculation process is not considered, and the fact that changes in overflow turbidity and multi-factor disturbances, such as temperature and pH fluctuations, can only be observed 15-30 minutes after the addition of the agent.
[0079] Therefore, the control strategy design of this invention follows the logic of first analyzing the coupling relationship, then matching the model architecture, and finally coordinating with the control device in real time for allocation:
[0080] First, statistical analysis of a large amount of experimental data revealed a nonlinear relationship between feed concentration (X2), flow rate (X3), and liquid level in the speed regulating tank (X1) (e.g., as the flow rate increases, the liquid level rises, which intensifies the circulation intensity). Furthermore, the three control parameters—flocculator dosage (Y1), addition location (Y2), and dilution water gate opening degree (Y3)—must simultaneously respond to feedback from overflow turbidity (Z1) and underflow concentration (Z2). In other words, adjusting a single control parameter cannot independently resolve the deviation of a certain indicator, and a multi-input-multi-output coupled model must be established.
[0081] Therefore, considering the complex coupling relationships, limited sample data, and large time delays and disturbances, traditional linear regression models and neural network models were ruled out, and the support vector machine regression model was finally selected. Its kernel function can effectively fit nonlinear relationships and can still maintain high prediction accuracy in small sample scenarios, making it suitable for the actual conditions of industrial field data collection.
[0082] To further improve the model's adaptability to disturbances, we consider using grid search, random search, or Bayesian optimization methods to precisely tune the penalty factor C (balancing fitting accuracy and generalization ability) and kernel coefficient γ (regulating the strength of nonlinear mapping) of the support vector machine, avoiding overfitting or underfitting. For the large time delay characteristic, we design a tiered feedback adjustment logic: when the deviation is ≤5%, we only fine-tune the model parameters, such as adjusting the penalty factor C, to quickly respond to short-term fluctuations; when the deviation is >10%, we re-collect a large amount of real-time data for model retraining to correct long-term coupling shifts and ensure the adjustment strategy remains effective under dynamic conditions.
[0083] In summary, the flocculant precision dosing control system and device of the present invention have significant innovation and practicality, and can effectively solve many problems such as the unsatisfactory dilution and flocculation effect of the feed well of the self-contained thickener. It is of great significance to promote the development of intelligent control technology in mineral processing plants.
[0084] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.
Claims
1. A thickener feed well dilution and flocculation control device, characterized in that, The system includes a slurry inlet pipe (1), a diversion speed regulating tank (2), and a main feed well (3). The slurry inlet pipe (1) enters the diversion speed regulating tank (2) tangentially. Multiple diversion branch pipes (21) are evenly arranged at the bottom of the diversion speed regulating tank (2). Multiple diversion branch pipes (21) extend into the main feed well (3). A liquid level difference strengthening baffle plate (31) is set at the height of the upper part of the main feed well (3) close to the liquid level, and multiple self-dilution water channels (4) are opened on the side wall in the circumferential direction. A flocculant solution adding ring pipe (5) is set in the annular zone between the outer side of the diversion speed regulating tank (2) and the inner side of the main feed well (3). The flocculant solution adding ring pipe (5) is located in the middle position above the liquid level difference strengthening baffle plate (31). Multiple flocculant solution nozzles (51) are evenly arranged in the circumferential direction at the lower part of the flocculant solution adding ring pipe (5). The outside of the self-dilution water channel (4) is connected to the overflow water of the thickener, and a dilution water volume regulating gate (41) and a drive connecting rod (42) are set at the outer end. The inner wall of the main feed well (3) is provided with an annular baffle (32), and the discharge ports at the ends of the multiple branch pipes (21) are located between the liquid level difference strengthening baffle (31) and the annular baffle (32); The liquid level difference enhanced turbulence plate (31) is composed of multiple inclined plates, which are evenly arranged in the circumferential direction inside the feed well and form an angle of 10-25° with the tangent of the slurry circulation direction. At the same time, it ensures that its top is located 10-30mm below the self-dilution water channel (4). The spray angle of the flocculant solution nozzle (51) is 45-60°, facing the central axis of the main feed well (3). Multiple flocculant solution nozzles (51) can switch between multiple sets of different position combinations according to the real-time data of feed flow rate and concentration.
2. The thickener feed well dilution and flocculation control device according to claim 1, characterized in that, The liquid level difference enhanced turbulence plate (31) consists of 6-12 inclined plates, and the vertical projection width is 1 / 8-1 / 5 of the diameter of the feed well.
3. A thickener feed well dilution and flocculation control system, comprising the thickener feed well dilution and flocculation control device as described in any one of claims 1-2, characterized in that, Also includes: The data acquisition layer, model prediction layer, operation execution layer, and feedback adjustment layer are all included.
4. The thickener feed well dilution and flocculation control system according to claim 3, characterized in that, The data acquisition layer includes a flow meter and a concentration meter installed on the slurry inlet pipe (1) and a level gauge installed above the flow-dividing speed regulating tank (2). The operation execution layer includes a flocculant solution addition loop pipe (5) whose front end is connected to a flocculant addition pump with controllable flow rate. Multiple flocculant solution nozzles (51) are connected to the control system and each nozzle can be controlled individually. The drive connecting rod (42) above each dilution water volume regulating gate (41) is connected to the power drive device. The feedback adjustment layer includes a turbidity meter installed on the overflow of the thickener and a concentration meter installed on the underflow.
5. The thickener feed well dilution and flocculation control system according to claim 4, characterized in that, It includes the following four steps: data collection, model prediction, operation execution, and feedback adjustment: S1: Data Acquisition: The flow rate, concentration and level of the feed slurry are collected using flow meters, concentration meters and level meters in the data acquisition layer. The turbidity and concentration meters in the feedback adjustment layer are used to collect the overflow turbidity and underflow concentration. The data is cleaned, missing values and outliers are processed, and the data is preprocessed using standardization methods to make each feature have the same scale. S2: Model prediction: Using machine learning methods, the optimal mathematical model that matches the results is pre-established based on experimental data, and the thresholds of each parameter of the prediction model are adjusted. Based on the calculation results of the prediction parameters, the self-dilution water volume, flocculant solution addition amount and addition location are set respectively. S3: Operation Execution: Based on the prediction results, the control execution module sends control signals to the relevant execution mechanisms to adjust the dilution water volume, the location and amount of flocculant added; The appropriate self-dilution water volume is set by controlling the opening and closing degree of the dilution water regulating gate; The flocculant addition amount is precisely controlled by a flocculant addition pump, and the appropriate addition position is selected by controlling the flocculant solution nozzle switch at different locations. S4: Feedback Adjustment: The monitoring and feedback module monitors the actual overflow turbidity and underflow concentration in real time, compares them with the predicted values, and if the deviation exceeds the set threshold, collects data again, updates and trains the model, and optimizes the control strategy.
6. The thickener feed well dilution and flocculation control system according to claim 5, characterized in that, In the feedback adjustment step, different deviation threshold levels are set, and different feedback and optimization measures are taken according to the severity of the deviation. For example, when there is a slight deviation, only the model parameters are fine-tuned, and when there is a severe deviation, a large amount of data is collected again to fully train the model.