Intelligent control method, system, medium and equipment for grinding process

Abstract

The present application provides a method, system, medium and equipment for intelligent control of a grinding process, the method comprising: obtaining real-time detection data of a grinding process loop, online detection data of grinding sound and frequency, and online detection data of ore size in real time; based on the real-time detection data, historical data, online analysis value of the filling rate of the grinding machine, and online analysis value of the ore size distribution, according to the reasoning logic of a pre-constructed dynamic expert knowledge base, real-time reasoning and identification of working condition characteristics are performed to obtain the current working condition characteristics, and the corresponding dynamic expert rules are determined according to the current working condition characteristics; based on the dynamic expert rules and the real-time detection data, the online analysis value of the filling rate of the grinding machine and the online analysis value of the ore size distribution are used as inputs of the intelligent process control algorithm to obtain the output vector of the dynamic characteristic model of the grinding process loop; and the grinding process loop is closed-loop controlled according to the output vector. The present application can improve the efficiency and accuracy of grinding.

Description

Technical Field

[0001] The present application relates to the field of grinding technology, and in particular to a method, system, medium and equipment for intelligent control of grinding process. Background Art

[0002] Grinding is to gradually reduce the ore particle size by using the impact and grinding effect of media (such as steel balls, steel rods, gravel) and the ore itself to meet the requirements of subsequent operations. In the ore dressing process, the main purpose of grinding is to separate the useful minerals from the gangue minerals as much as possible to meet the needs of the subsequent separation process.

[0003] The grinding and classification process is an important part of the mineral processing process, and its purpose is to grind the ore into a particle size that meets the requirements of the next mineral processing step. Among them, the SABC process (closed-circuit grinding and classification process of "semi-autogenous grinding mill + ball mill + stone crusher") has gradually become the main development direction of the grinding process in the mineral processing process due to its advantages such as large ore processing capacity, small footprint, streamlined equipment, low production cost, and low maintenance intensity. The SABC process can be divided into a semi-autogenous grinding-stone crushing circuit and a ball mill-cyclone classification circuit. Among them, the control effect of the semi-autogenous grinding-stone crushing circuit directly determines the efficiency of the entire SABC process. It is imperative to achieve the optimized operation and green production of the semi-autogenous grinding-stone crushing circuit. However, this circuit has large disturbance to the ore, complex working condition control, and key process indicators are difficult to detect online. There are many influencing factors, etc. It is extremely challenging to study the intelligent control technology of the semi-autogenous grinding-stone crushing circuit. Summary of the invention

[0004] In view of this, the present application provides a method, system, medium and equipment for intelligent control of a grinding process, the main purpose of which is to improve the grinding efficiency and precision through intelligent control of the grinding process.

[0005] According to one aspect of the present application, a grinding process intelligent control method is provided, which is used to intelligently control a grinding process circuit, wherein the grinding process circuit refers to a circuit consisting of at least a grinding machine and a crusher, and the method comprises:

[0006] Real-time acquisition of real-time detection data of the grinding process circuit, online detection data of grinding sound and frequency, and online detection data of ore block size. The real-time detection data includes operating status parameters and equipment parameters of the grinding process circuit;

[0007] Based on the real-time detection data, the historical data obtained from the pre-established historical database, the online analysis value of the mill filling rate obtained by analyzing the grinding sound and grinding frequency online detection data, and the online analysis value of the ore fragmentation distribution obtained by analyzing the ore fragmentation online detection data, according to the reasoning logic of the pre-established dynamic expert knowledge base, real-time working condition characteristic reasoning and identification are performed to obtain the current working condition characteristic, and the corresponding dynamic expert rules are determined according to the current working condition characteristic;

[0008] Based on the dynamic expert rules, the real-time detection data, the online analysis value of the mill filling rate and the online analysis value of the ore size distribution as inputs of the intelligent process control algorithm, an output vector of the dynamic characteristic model of the grinding process loop is obtained;

[0009] The grinding process loop is closed-loop controlled according to the output vector.

[0010] In one implementation, the real-time operating condition characteristic reasoning and identification includes:

[0011] Based on the real-time detection data, the historical data, the online analysis value of the mill filling rate, and the online analysis value of the ore size distribution, at least one of the axial pressure, power, axial pressure change rate, power change rate, and filling rate is determined as an identification index to perform real-time operating condition characteristic reasoning and identification.

[0012] In one implementation, determining at least one of shaft pressure, power, shaft pressure change rate, power change rate, and filling rate as the identification indicator includes:

[0013] Determining at least one of an axial pressure setting value, a power setting value, an axial pressure change rate setting value, a power change rate setting value, and a filling rate setting value; and,

[0014] Determine at least one of an actual value of axial pressure, an actual value of power, an actual value of axial pressure change rate, an actual value of power change rate, and an actual value of filling rate according to an output vector of the dynamic characteristic model;

[0015] The identification index is determined by combining the set value and the actual value.

[0016] In one implementation, the method further includes:

[0017] Based on the dynamic expert rules, the dynamic expert knowledge base is updated in real time, wherein the dynamic expert knowledge base is used to store and manage expert knowledge of grinding process circuit operating condition identification and operating experience, so as to identify the operating condition type according to the expert knowledge and give corresponding expert rules, and the expert rules include at least one of the control state, control sequence, change amplitude and duration of the intelligent control vector of the dynamic characteristic model.

[0018] In one implementation, determining the corresponding dynamic expert rule according to the current operating condition characteristics includes:

[0019] Determine the current operating condition characteristic identification result, wherein the identification result includes at least one of a control state, a control sequence, a change amplitude, and a duration of the intelligent control vector to be generated;

[0020] According to at least one of the control state, control sequence, change amplitude and duration, an expert control vector of the dynamic expert rule is calculated.

[0021] In one implementation, the method further includes:

[0022] An initial model of the dynamic characteristic model is established, wherein the expert control vector of the dynamic expert rule is applied to the controlled vector of the dynamic characteristic model, and a dynamic transfer function matrix describing the dynamic characteristics of the grinding process loop is established based on the expert control vector and the controlled vector to obtain an initial model.

[0023] In one implementation, the method further includes: solving the model parameters of the dynamic characteristic model by the following steps:

[0024] a) Set the validity period of the dynamic characteristic model, linearize the model within the validity period, and determine that the dynamic characteristic model is a linear steady-state system within this period;

[0025] b) For a linear steady-state system, the dynamic characteristic model is represented based on a dynamic transfer function matrix and an expert control vector: Yo(s)=G(s)V(s), wherein Yo(s) represents the controlled vector, G(s) represents the dynamic transfer function, and V(s) represents the expert control vector.

[0026] In one implementation, the dynamic characteristic model is a model used to describe the relationship between the intelligent control vector and the controlled vector, wherein the intelligent control vector includes at least one of the feed rate, the amount of water added, the rotation rate, the feed block size, and the amount of steel balls added, and the controlled vector includes at least one of the power, the axial pressure, and the filling rate.

[0027] In one implementation, the control process of the intelligent process control algorithm is:

[0028] According to the transfer function matrix G(s) of the dynamic characteristic model, the intelligent controller D(s) is designed, and the intelligent control vector W(s) is expressed as: W(s) = D(s)V(s), where V(s) represents the expert control vector;

[0029] The parameters of the intelligent controller are solved by the following steps: W(s) is solved according to Yo(s)=G(s)W(s), where Yo(s) represents the controlled vector; D(s) is solved according to W(s)=D(s)V(s).

[0030] In one implementation, the method further includes: dynamically updating the dynamic characteristic model, wherein:

[0031] By setting the validity period of the dynamic characteristic model, when the error between the predicted output value and the actual output value is greater than a certain threshold, it is confirmed that the dynamic characteristic model is invalid;

[0032] When the dynamic characteristic model fails, the parameters of the transfer function matrix of the dynamic characteristic model are solved according to the output vector and actual output value of the intelligent control algorithm in the previous validity period as the model parameters in the next validity period to dynamically update the dynamic characteristic model.

[0033] In one implementation, after dynamically updating the dynamic characteristic model, the method further includes:

[0034] Update the parameters of the intelligent controller D(s) according to the preset values ​​X(s) and the expert control vector V(s).

[0035] According to one aspect of the present application, a grinding process intelligent control system is provided, which is used to intelligently control a grinding process loop, wherein the grinding process loop refers to a loop consisting of at least a grinding machine and a crusher, and the system comprises:

[0036] A data acquisition module is used to acquire real-time detection data of the grinding process circuit, online detection data of grinding sound and frequency, and online detection data of ore block size. The real-time detection data includes operating status parameters and equipment parameters of the grinding process circuit;

[0037] The working condition identification and expert rule determination module is used to perform real-time working condition characteristic reasoning and identification according to the reasoning logic of the pre-constructed dynamic expert knowledge base based on the real-time detection data, the historical data obtained from the pre-established historical database, the online analysis value of the mill filling rate obtained by analyzing the online detection data of the grinding sound and grinding frequency, and the online analysis value of the ore fragmentation distribution obtained by analyzing the online detection data of the ore fragmentation, to obtain the current working condition characteristics, and determine the corresponding dynamic expert rules according to the current working condition characteristics;

[0038] A model prediction module is used to obtain an output vector of a dynamic characteristic model of a grinding process loop based on the dynamic expert rules, the real-time detection data, the online analysis value of the mill filling rate, and the online analysis value of the ore size distribution as inputs of an intelligent process control algorithm;

[0039] A control execution module is used to perform closed-loop control on the grinding process loop according to the output vector.

[0040] According to one aspect of the present application, a storage medium is provided, in which a computer program is stored, wherein the computer program is configured to execute the above method when running.

[0041] According to one aspect of the present application, an electronic device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the above method.

[0042] Through the above-mentioned technical scheme, the present application provides an intelligent control method, system, medium and equipment for grinding process, which performs real-time reasoning and identification of operating condition characteristics based on real-time detection data, historical data, online analysis values ​​of grinding machine filling rate and online analysis values ​​of ore size distribution according to the reasoning logic of a pre-constructed dynamic expert knowledge base, obtains the current operating condition characteristics, and determines the corresponding dynamic expert rules according to the current operating condition characteristics; based on dynamic expert rules and real-time detection data, the online analysis values ​​of grinding machine filling rate and online analysis values ​​of ore size distribution are used as inputs of the intelligent process control algorithm to obtain the output vector of the dynamic characteristic model of the grinding process loop; and the grinding process loop is closed-loop controlled according to the output vector. It can be seen that the present application realizes real-time and intelligent control of the grinding process loop through intelligent process control algorithms and dynamic characteristic models, thereby improving the grinding efficiency. For example, by feeding back the results of model prediction to the operating condition reasoning and identification, the operating condition identification and dynamic expert rule determination and update are realized according to the latest controlled vector results (such as power, axial pressure, filling rate); and the online analysis values ​​of the mill filling rate and the online analysis values ​​of the ore block size distribution are integrated into the operating condition characteristic reasoning and identification as important process indicators, which increases the operating condition description dimension and improves the accuracy of the operating condition characteristic reasoning and identification.

[0043] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0045] Picture 1 The flowchart of the intelligent control method of the grinding process provided by the embodiment of the present application is shown;

[0046] Picture 2 A schematic diagram of an intelligent control system for a grinding process provided in an embodiment of the present application is shown;

[0047] Picture 3 A schematic diagram of an operating condition characteristic reasoning and identification process provided by an embodiment of the present application is shown;

[0048] Picture 4 A schematic diagram of the principle of an intelligent control system for a semi-autogenous grinding-rock crushing circuit provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0049] In order to enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only embodiments of a part of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work should fall within the scope of protection of the present application. It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict.

[0050] The inventors found in the research process that the existing control method for the SAG mill-rock crushing circuit in the SABC process mainly adopts an expert control strategy, that is, by collecting a large amount of operating experience of on-site operators and dispatchers, summarizing it into expert control rules, and treating the SAG mill-rock crushing circuit as a black box model. Based on fuzzy reasoning, the value of the control vector is obtained to achieve the control of the SAG mill-rock crushing circuit. In addition, there are also control methods based on SAG mill simulation models, which are mainly based on mechanism models and process models, including overall balance models, material distribution models, and discrete element models.

[0051] The control method based on the expert control strategy does not rely on the mechanism model or process model of the semi-autogenous grinding-stone crushing circuit. It can judge the working conditions based on the expert control rules and use the fuzzy reasoning method to obtain the control vector value. The disadvantage is that when the working conditions change significantly or the working conditions outside the expert control rules appear, it is difficult for this method to obtain the control vector through the existing rules, that is, this method has a strong dependence on the expert rules. At the same time, the collection and summary of expert rules often takes a lot of time, and the adaptability of expert control rules of different concentrators is also poor, and the expert control rules need to be collected and summarized again. On the other hand, when the control vector acts directly on the circuit, there are usually problems of weak control stability, low accuracy, and poor rapidity.

[0052] The control method based on the SAG mill simulation model is to obtain the material balance relationship of the SAG mill-rock crushing circuit through mechanism and process analysis, simulate the material distribution and movement trajectory inside the SAG mill, analyze the change law of the SAG mill liner, steel balls, and discharge grille when the control vector changes, and calculate the control vector value based on the analysis results. The disadvantage is that this method is highly dependent on the SAG mill simulation model, and the actual working conditions have many interference factors and are mutually coupled. At the same time, the parameters of the SAG mill may change with the working conditions. At this time, the accuracy of the simulation model in describing the actual working conditions will be reduced, and the control vector value obtained based on the simulation model is difficult to effectively control the SAG mill-rock crushing circuit. At the same time, these models focus on the change law of the parameters of the simulation equipment itself and the distribution trajectory and movement law of the materials, but they are also insufficient in describing the dynamic characteristics of the circuit. It is difficult to describe the relationship between the control vector (SAG mill feed rate, water addition, rotation rate, feed block size, steel ball addition) and the controlled vector (SAG mill power, shaft pressure, filling rate), and it is difficult to directly participate in the circuit control.

[0053] At the same time, in recent years, with the rapid development of detection and analysis technology, new detection instruments and analysis equipment for SAG-rock crushing circuits have been developed and applied, mainly including SAG feed block size analyzer, SAG mill cylinder vibration and strain detection equipment, and SAG mill cylinder material drop point audio detector. The SAG feed block size analyzer can obtain the statistical analysis value of the SAG feed block size distribution, which can be used to analyze the grindability of the ore. The SAG mill cylinder vibration and strain detection equipment can obtain the vibration and strain curves of the cylinder during the rotation process, which can be used to analyze the starting point and drop point of the material in the cylinder. The SAG mill cylinder material drop point audio detector can obtain the steel ball drop point audio curve, which is used to analyze whether the material drop point is within a reasonable range. However, the detection and analysis results of these new detection instruments and analysis equipment are currently only used as reference information for operators to analyze working conditions and equipment status, and have not yet participated in the automatic control of the SAG-rock crushing circuit.

[0054] In view of the above problems, the present application proposes an intelligent control solution for grinding process, aiming to solve the following technical problems:

[0055] (1) Solve the problem that it is difficult to obtain an effective control vector through existing rules when the operating conditions change significantly or when operating conditions outside the expert control rules appear;

[0056] (2) Solve the problem of low accuracy of simulation models in describing actual working conditions;

[0057] (3) Solve the problem that the existing mechanism and process model are difficult to describe the relationship between the control vector (SAG mill feed rate, water addition, rotation rate, feed block size, and steel ball addition) and the controlled vector (SAG mill power, shaft pressure, and filling rate);

[0058] (4) Solve the problem of how to use the data and information obtained by new detection instruments and analysis equipment to participate in loop control;

[0059] (5) Solve the problems of poor stability, low accuracy and poor speed when controlling the vector direct control loop.

[0060] It can be understood that although the embodiment of the present application is proposed from the problem of intelligent control of the semi-autogenous grinding-stone crushing circuit, the solution can be applied to other grinding process circuits in addition to the grinding process circuit of the semi-autogenous grinding-stone crushing circuit, such as the autogenous grinding-stone crushing circuit, the ball mill-stone crushing circuit, the rod mill-stone crushing circuit, etc. For convenience, this article mainly takes the grinding process circuit of the semi-autogenous grinding-stone crushing circuit as an example to illustrate how to intelligently control the grinding process circuit.

[0061] See also Picture 1 , shows a flow chart of the grinding process intelligent control method provided by the embodiment of the present application. The grinding process intelligent control method is used to intelligently control the grinding process loop, wherein the grinding process loop refers to a loop composed of at least a grinding machine and a crusher, and specifically includes the following steps:

[0062] S101: acquiring real-time detection data of the grinding process circuit, online detection data of grinding sound and frequency, and online detection data of ore block size. The real-time detection data includes operating status parameters and equipment parameters of the grinding process circuit.

[0063] Among them, the operating status parameters can be understood as the operating parameters of the grinding process circuit. Taking the semi-autogenous grinding-stone crushing circuit as an example, the operating status parameters can be the semi-autogenous grinding machine circuit detection process parameters, such as semi-autogenous grinding feed rate, semi-autogenous grinding water addition rate, stone return rate, stone bin material rate, etc. The equipment parameters refer to the equipment parameters of the grinding machine and / or crusher in the grinding process circuit. Still taking the semi-autogenous grinding-stone crushing circuit as an example, the equipment parameters can be the semi-autogenous grinding machine shaft pressure, automatic ball adding machine weight parameters, etc.

[0064] In addition to conventional real-time detection data, the embodiments of the present application also specifically obtain online detection data of grinding sound and grinding frequency and online detection data of ore block size as the basis for subsequent working condition identification, which can improve the identification dimension and accuracy. Among them, the grinding sound and grinding frequency can be detected in real time through a semi-autogenous grinding sound (electric ear) and grinding frequency (cylinder vibration and strain sensor) online detection unit. Similarly, the ore block size data can be detected in real time through an online ore block size detection unit.

[0065] After detecting the online detection data of grinding sound and grinding frequency and the online detection data of ore piece size, the online detection data of grinding sound and grinding frequency and the online detection data of ore piece size are analyzed to obtain the online analysis values ​​of grinding mill filling rate and the online analysis values ​​of ore piece size distribution. Therefore, in the subsequent working condition identification link, the online analysis values ​​of grinding mill filling rate and the online analysis values ​​of ore piece size distribution are integrated into the working condition characteristic reasoning and identification as important process indicators, which increases the working condition description dimension and improves the accuracy of working condition characteristic reasoning and identification.

[0066] S102: Based on real-time detection data, historical data obtained from a pre-established historical database, online analysis values ​​of mill filling rate obtained through analysis of online detection data of grinding sound and grinding frequency, and online analysis values ​​of ore fragmentation distribution obtained through analysis of online detection data of ore fragmentation, real-time operating condition characteristic reasoning and identification are performed according to the reasoning logic of a pre-constructed dynamic expert knowledge base to obtain the current operating condition characteristics, and corresponding dynamic expert rules are determined according to the current operating condition characteristics.

[0067] Among them, the dynamic expert knowledge base is used to store and manage expert knowledge of grinding process circuit working condition identification and operating experience, so as to identify the working condition type according to the expert knowledge and give corresponding expert rules, and the expert rules include at least one of the control state, control sequence, change amplitude, and duration of the intelligent control vector of the dynamic characteristic model mentioned later. In the embodiment of the present application, the dynamic expert knowledge base can be dynamically updated. For example, when the working condition changes significantly or a working condition outside the expert control rules occurs, effective control of the working condition can be achieved by adding expert knowledge and operating rules online.

[0068] In a specific implementation, a working condition characteristic reasoning and identification system can be used to identify working conditions and determine the expert rules corresponding to the working conditions. In one method, shaft pressure, power, shaft pressure change rate, power change rate, and filling rate are determined as indicators and input into the working condition characteristic reasoning and identification system to perform real-time working condition characteristic reasoning and identification according to the reasoning logic of the pre-established dynamic expert knowledge base (i.e., the strategy for regulating the control state, control sequence, change amplitude, and duration of the intelligent control vector of the dynamic characteristic model), where the working condition characteristic can be understood as the actual working condition level, such as the first level, the second level, etc., or understood as normal working condition, abnormal working condition, etc. After determining the operating condition characteristics, the output vector of the identification system (the control state, control sequence, change amplitude, and duration of the intelligent control vector to be generated) is inferred based on the operating condition characteristics to determine the expert control vector corresponding to the current operating condition, where the expert control vector can be a vector composed of feed amount, feed block size, added water amount, rotation rate, and steel ball addition amount. The expert control vector is used as the input of the intelligent process control algorithm in the subsequent steps, together with other inputs (real-time detection data, online analysis values ​​of mill filling rate, and online analysis values ​​of ore block size distribution) as the basis for intelligent process control, and intelligent process control prediction is performed.

[0069] As mentioned above, axial pressure, power, axial pressure change rate, power change rate, and filling rate can be determined as indicators to perform inference and identification of working condition characteristics. Therefore, in one implementation, according to real-time detection data, historical data, online analysis value of mill filling rate, and online analysis value of ore block size distribution, at least one of axial pressure, power, axial pressure change rate, power change rate, and filling rate is determined as an identification indicator to perform real-time inference and identification of working condition characteristics. In particular, it can be implemented in the following way: first, at least one of the axial pressure set value, power set value, axial pressure change rate set value, power change rate set value, and filling rate set value is determined; then, according to the output vector of the dynamic characteristic model, at least one of the axial pressure actual value, power actual value, axial pressure change rate actual value, power change rate actual value, and filling rate actual value is determined; finally, the set value and the actual value are combined to determine the comprehensive value of each indicator as the final identification indicator.

[0070] S103: Based on dynamic expert rules and real-time detection data, the online analysis value of the mill filling rate and the online analysis value of the ore size distribution are used as inputs of the intelligent process control algorithm to obtain the output vector of the dynamic characteristic model of the grinding process loop.

[0071] In addition to the general existing grinding process system control which only performs process control, the embodiment of the present application adds an intelligent control method. The so-called intelligent control can be understood as the real-time and intelligent control of the grinding process loop through the intelligent process control algorithm and the dynamic characteristic model. For example, the results of model prediction are fed back to the working condition reasoning identification, thereby realizing the working condition identification and dynamic expert rule determination and updating according to the latest controlled vector results (such as power, axial pressure, filling rate). Among them, the intelligent process control algorithm is based on a new control method predictive control that has been independently developed in the process of industrial practice. Practical predictive control methods include dynamic matrix control (DMC), model algorithm control (MAC), generalized predictive control (GPC), model predictive heuristic control (MPHC) and predictive function control (PFC). The loop dynamic characteristic model is a model that describes the relationship between the intelligent control vector (such as feed rate, additional water amount, rotation rate, feed block size, steel ball addition amount) and the controlled vector (such as power, axial pressure, filling rate). When modeling, data dimensionality reduction can be performed first, and then a multi-input and multi-output model can be established using a nonlinear correlation analysis method. The principles and implementation of the intelligent process control algorithm and dynamic characteristic model will be described in detail below.

[0072] In addition, the dynamic characteristic model can be dynamically updated. For example, the validity period of the dynamic characteristic model is set. When the error between the predicted output value and the actual output value is greater than a certain threshold, the dynamic characteristic model becomes invalid and the dynamic characteristic model and the corresponding intelligent controller need to be re-obtained. When the dynamic characteristic model fails, the parameters of the dynamic characteristic model transfer function matrix G (s) are solved according to the output vector and actual output value of the intelligent control algorithm in the previous validity period, and used as the model parameters in the next validity period to realize dynamic update of the model.

[0073] S104: Performing closed-loop control on the grinding process circuit according to the output vector.

[0074] Therefore, the grinding process intelligent control method provided in the embodiment of the present application solves the problem that it is difficult to obtain an effective control vector through existing rules when the working conditions change significantly or working conditions outside the expert control rules appear by setting and updating a dynamic expert knowledge base; and models the controlled object according to production data and process mechanisms to solve the problem of low accuracy of the simulation model when describing the actual working conditions; and directly uses the control vectors (semi-autogenous mill feed rate, water addition, rotation rate, feed block size, steel ball addition) and controlled vectors (semi-autogenous mill power, shaft pressure, filling rate) in the production process to model, solving The problem that the existing mechanism and process model are difficult to describe the relationship between the control vector and the controlled vector is solved; in addition, the online analysis values ​​of the mill filling rate and the online analysis values ​​of the ore block size distribution are integrated into the reasoning and identification of the operating condition characteristics as important process indicators, which increases the dimension of the operating condition description, improves the accuracy of the reasoning and identification of the operating condition characteristics, and solves the problem of how the data and information obtained by the new detection instruments and analysis equipment participate in the loop control; in addition, an intelligent process control algorithm and dynamic update method are designed for the grinding process loop to solve the problems of weak stability, low accuracy and poor speed when the control vector directly controls the loop.

[0075] Below, the intelligent control scheme of the semi-autogenous grinding-hard stone crushing circuit is taken as an example to explain the embodiment of the present application in detail. It can be understood that in this example, the grinding machine specifically refers to the semi-autogenous grinding machine, and accordingly, the online analysis value of the grinding machine filling rate specifically refers to the online analysis value of the semi-autogenous grinding machine filling rate, and so on.

[0076] See also Picture 2 , which is a schematic diagram of the intelligent control system of the semi-autogenous grinding-stone crushing circuit.

[0077] The system consists of three parts: intelligent perception and execution layer, advanced process control layer, and expert analysis and decision-making layer.

[0078] (1) Intelligent perception and execution layer

[0079] The intelligent perception and execution layer consists of process control intelligent detection unit, equipment intelligent monitoring unit, intelligent actuator, ore size online detection unit, semi-autogenous grinding sound (electric ear) grinding frequency (drum vibration and strain sensor) online detection unit. Among them, the ore size online detection unit and the semi-autogenous grinding sound (electric ear) grinding frequency (drum vibration and strain sensor) online detection unit belong to the semi-autogenous grinding-hard stone crushing circuit intelligent control method and device component module.

[0080] The process control intelligent detection unit refers to the sensor / transmitter used to detect process parameters in the SAG mill circuit, such as the SAG mill feed belt scale, SAG mill water addition flowmeter, stone return belt scale, stone bin level meter, etc.; the equipment intelligent monitoring unit refers to the sensor / transmitter and control unit that comes with the SAG mill circuit equipment, such as the SAG mill shaft pressure sensor, automatic ball adding machine weighing meter, etc.; the intelligent actuator refers to the actuator of the control vector of the SAG mill circuit, such as the heavy plate feeder inverter, SAG mill feed belt inverter, SAG mill inverter, SAG mill water addition regulating valve; the ore size online detection unit is used to obtain real-time high-precision image information of the SAG mill feed ore size, the SAG mill sound (electric ear) is used to obtain the audio signal of the SAG material landing point, and the SAG mill frequency (cylinder vibration and strain sensor) is used to detect the vibration and strain signals of the cylinder online.

[0081] (2) Advanced process control layer

[0082] The advanced process control layer is composed of a process control system (such as a distributed control system DCS, a bus control system FCS, etc.) platform and an intelligent control system platform. Among them, the process control system is composed of a real-time detection data acquisition and transmission unit and a control vector output unit. The process control system platform is based on I / O cards and communication cards to collect and transmit production process data in real time; the intelligent control system platform belongs to the semi-autogenous grinding-rock crushing circuit intelligent control method and device component module, which is composed of an intelligent process control algorithm and a semi-autogenous grinding-rock crushing circuit dynamic characteristic model.

[0083] The intelligent process control algorithm is based on predictive control, a new control method independently developed in the process of industrial practice. Practical predictive control methods include dynamic matrix control (DMC), model algorithm control (MAC), generalized predictive control (GPC), model predictive heuristic control (MPHC) and predictive function control (PFC). The dynamic characteristic model of the semi-autogenous mill-rock crushing circuit is a model that describes the relationship between the control vector (semi-autogenous mill feed rate, water addition, speed, feed block size, steel ball addition) and the controlled vector (semi-autogenous mill power, shaft pressure, filling rate). When modeling, data dimensionality reduction is first performed, and then a multi-input and multi-output model is established using a nonlinear correlation analysis method.

[0084] Advanced process control integrates the dynamic characteristic model of the semi-autogenous grinding-rock crushing circuit on the process control system software platform and embeds the intelligent process control algorithm to achieve real-time collection of production process data and real-time control of process parameter indicators. At the same time, the intelligent control system can display the control effect in the process control system in the form of configuration, realizing the interconnection of the two systems.

[0085] (3) Expert analysis and decision-making level

[0086] The expert analysis decision layer consists of a dynamic expert knowledge base, a historical database, a working condition characteristic reasoning and identification system, an ore size analysis module, and a semi-autogenous mill filling rate analysis module. Each module in this part belongs to the semi-autogenous mill-hard rock crushing circuit intelligent control method and device.

[0087] The dynamic expert knowledge base is used to store and manage expert knowledge of semi-autogenous grinding-rock crushing circuit working condition identification and operating experience, identify working condition types based on expert knowledge and give corresponding operating rules. This knowledge base is dynamic, that is, expert knowledge can be added in real time; the historical database is used to store production process data throughout the life cycle, providing data support for experts to identify working condition types; the ore size analysis module analyzes and counts image information based on machine vision perception algorithms to obtain real-time distribution and change information of ore size; the semi-autogenous grinding mill filling rate online analysis module is based on the audio signal of the semi-autogenous grinding material landing point obtained by grinding sound (electric ear), and the vibration and strain signal of the cylinder detected by the semi-autogenous grinding frequency (cylinder vibration and strain sensor), comprehensively analyzes and determines the starting and landing points of the semi-autogenous grinding mill materials, and then calculates the semi-autogenous grinding mill filling rate.

[0088] The operating condition characteristics reasoning and identification system comprehensively analyzes historical data, real-time detection data, real-time distribution and change information of ore block size, online analysis value of semi-autogenous mill filling rate, and equipment status information. It identifies the operating condition characteristics based on the dynamic expert knowledge base, and then obtains the corresponding expert rules based on the operating condition characteristics. Similarly, the expert rules are also dynamic, that is, the rules can be modified in real time. The operating condition characteristics reasoning and identification process is as follows: Picture 3 shown.

[0089] The information flow in the overall structure of the intelligent control system of the semi-autogenous grinding-rock crushing circuit is:

[0090] a) The signals of the process control intelligent detection unit and equipment intelligent monitoring unit in the intelligent perception and execution layer are sent to the process control system through the real-time detection data acquisition and transmission module in the advanced control layer, and then sent to the working condition characteristic reasoning and identification system in the expert analysis and decision-making layer through OPC communication, as one input of the working condition characteristic reasoning and identification system;

[0091] b) The operating condition characteristics reasoning and identification system comprehensively analyzes historical data, real-time detection data, real-time distribution and change information of ore block size, and online analysis value of semi-autogenous mill filling rate, and identifies the operating condition characteristics based on the dynamic expert knowledge base, and then obtains the corresponding expert rules according to the operating condition characteristics;

[0092] c) Taking expert rules, real-time detection data, online analysis value of SAG mill filling rate, and online analysis information of SAG mill feed size distribution as input, based on intelligent process control algorithm, the output value of SAG mill-rock crushing circuit dynamic characteristic model is obtained;

[0093] d) The output vector is transmitted to the intelligent actuator and the actuator equipped with the intelligent monitoring unit through the control vector output unit of the process control system to realize the closed-loop control of the semi-autogenous grinding-rock crushing circuit.

[0094] like Picture 4 , which is a schematic diagram of the principle of the intelligent control system of the semi-autogenous grinding-stone crushing circuit.

[0095] Picture 4 The vectors of the semi-autogenous grinding-rock crushing intelligent control system are as follows:

[0096] a) Set value vector X =[ x 1, x 2, x 3. x 4. x 5] T , that is, the axial pressure of the semi-autogenous mill x 1. Power x 2. Axial pressure change rate x 3. Power change rate x 4. Filling rate x 5;

[0097] b) The feedback value (actual value / analyzed value) vector is Y F =[ y 1, y 2, y 3. y 4. y 5] T , that is, the axial pressure of the semi-autogenous mill y 1. Power y 2. Axial pressure change rate y 3. Power change rate y 4. Filling rate y 5;

[0098] c) The input vector of the operating condition characteristic reasoning and identification system is: Z =[ z 1, z 2, z 3. z 4. z 5. z 6] T and other knowledge information, among which, z i = x i - y i ( i =1, 2, ..., 5), other knowledge information includes dynamic expert knowledge base, historical database and equipment status information, etc.;

[0099] d) The output of the operating condition characteristic reasoning and identification system includes four parts: the control state of the control vector, the control sequence, the change amplitude, and the duration. K =[ k 1, k 2, k 3. k 4. k 5] T Refers to the state of each vector in the control vector, which can be divided into three states: positive action (increase), no action (unchanged), and negative action (decrease), represented by 1, 0, and -1 respectively; control order L =[ l 1, l 2, l 3. l 4. l 5] T Refers to the order of control of each vector in the control vector (the control order of the vectors can be the same), represented by 1, 2, ..., 5, Null (Null means that the vector does not participate in the control under this working condition); the change amplitude value U =[ u 1, u 2, u 3. u 4. u 5] T Duration T =[ t 1, t 2, t 3. t 4. t 5] T , in duration T According to the control sequence, the entire control process is divided into n Time period ( n Equal to the maximum value of the elements in the control sequence, such as the control sequence L =[1, 2, 3, 4, 5] T hour, n =5);

[0100] e) The expert control vector is V =[ v 1, v 2, v 3. v 4. v 5] T . That is, the feed rate of the semi-autogenous mill v 1. Feed block v 2. Add water v 3. Turn rate v 4. Steel ball addition amount v 5;V Depend on K , L , U and T Through operation, we can get: the maximum value of the elements in the control sequence n = j ( j =1, 2, ..., 5), the control sequence s i ≤ j ( i =1, 2, ..., 5) of the control vector k i =1 or -1, then z i = k i × u i , that is, in n In a period of time, z i = k i × u i .

[0101] f) Intelligent process control algorithm output vector, that is, intelligent control vector W =[ w 1, w 2, w 3. w 4. w 5] T . Expert Control Vector V =[ v 1, v 2, v 3. v 4. v 5] T , as the input of the intelligent process control algorithm, after calculation by the intelligent process control algorithm, the intelligent control vector is obtained W =[ w 1, w 2, w 3. w 4. w 5] T .

[0102] g) The controlled vector is Yo =[ y 1, y 2, y 5] T . That is, intelligent control vector W =[ w 1, w 2,w 3. w 4. w 5] T Acting on the dynamic characteristic model of the semi-autogenous grinding-rock crushing circuit, the controlled vector Yo =[ y 1, y 2, y 5] T , that is, the axial pressure of the semi-autogenous mill y 1. Power y 2. Filling rate y 5.

[0103] The dynamic characteristic model of the SAG-rock crushing circuit refers to the SAG mill feed rate. v 1. Feed block v 2. Add water v 3. Turn rate v 4. Amount of steel balls added v 5 is the control vector (i.e. the input is V =[ v 1, v 2, v 3. v 4. v 5] T ), based on the actual value of the semi-autogenous mill shaft pressure y 1. Actual power value y 2. Filling rate analysis value y 5 is the controlled vector (that is, the output is Yo =[ y 1, y 2, y 5] T ), a model that describes the dynamic characteristics of the process vectors of the semi-autogenous grinding-rock crushing circuit. The semi-autogenous grinding-rock crushing circuit has the characteristics of nonlinearity, strong coupling, and time-varying. The method of establishing the dynamic model of this circuit through mechanism or process analysis has great limitations.

[0104] In the embodiment of the present application, the model is linearized in time-sharing mode by utilizing the characteristics of the computer control system's strong online computing capability and discrete control. The model structure is determined by mechanism and process analysis. The model parameters are solved based on production process data analysis and system identification methods. The specific steps are as follows:

[0105] a) Set the validity period of the dynamic characteristic model, and linearize the model within the validity period, that is, time-sharing linearization. At the same time, assume that the loop model is steady-state (that is, non-time-varying system) within this time period, then the model is a linear steady-state system.

[0106] b) For linear steady-state systems, based on the transfer function of classical control theory (using G ( s) represents) represents the dynamic characteristic model of the semi-autogenous grinding-stone crushing circuit, then Yo ( s )= G ( s ) V ( s ),in, G ( s ) is the transfer function matrix of the dynamic characteristic model, we can get:

[0107]

[0108] and V =[ v 1, v 2, v 3. v 4. v 5] T , Yo =[ y 1, y 2, y 5] T , then:

[0109]

[0110] Thus, the dynamic characteristic model of the semi-autogenous grinding-stone crushing circuit is obtained.

[0111] As mentioned above, intelligent process control algorithms can be implemented through dynamic matrix control (DMC), model algorithm control (MAC), generalized predictive control (GPC), model predictive heuristic control (MPHC) and predictive function control (PFC). The general principles and implementation process are as follows:

[0112] Design of intelligent controller based on the transfer function matrix of the dynamic characteristic model of the semi-autogenous grinding-rock crushing circuit D ( s ),make:

[0113]

[0114] At this time, the intelligent control vector W ( s )for: W ( s )= D ( s ) V ( s ),Right now:

[0115]

[0116] W ( s) is used as the input of the dynamic characteristic model of the semi-autogenous grinding-rock crushing circuit. Yo ( s )= G ( s ) W ( s )= G ( s )[ D ( s ) V ( s )].

[0117] make Yo ( s ) is the set value Y ( s ),at this time G ( s ) is known, we can use Yo ( s )= G ( s ) W ( s ) Solution W ( s ),Right now

[0118] W ( s )= G -1 ( s ) Yo ( s )

[0119] and V ( s ) is also known, and can be based on W ( s )= D ( s ) V ( s ) Solution D ( s ),Right now

[0120] D ( s )= W ( s ) V -1 ( s ) = [ G -1 ( s ) Yo ( s )] V -1 ( s )

[0121] Through the above calculation method, the parameters of the intelligent process controller can be solved.

[0122] In addition, the dynamic characteristic model of the semi-autogenous grinding-rock crushing circuit can be dynamically updated. Dynamic characteristic model G ( s ) During the validity period, the model parameters remain unchanged, and the output vector of the intelligent control algorithm during the validity period of this period W and output value Yo , solve the transfer function matrix of the dynamic characteristic model G ( s ) as the model parameters in the next validity period to achieve dynamic model update. After the dynamic characteristic model of the semi-autogenous grinding-rock crushing circuit is dynamically updated, the set value Y ( s ) and the expert control vector V ( s ) Update the intelligent process controller D ( s ) parameters.

[0123] In a specific implementation, the embodiment of the present application can be implemented through the following steps:

[0124] (1) Establish a historical database, a dynamic expert knowledge base, and an operating condition characteristic reasoning and identification system, import historical data into the historical database, and write expert knowledge and operating rules into the dynamic expert knowledge base. The expert knowledge and operating rules can be written into the dynamic expert knowledge base in real time.

[0125] (2) Setting up an online analysis system for ore size and an online analysis system for SAG mill filling rate, and obtaining online statistical analysis values ​​of the ore size distribution of the SAG mill feed and the SAG mill filling rate analysis values;

[0126] (3) The real-time process detection data, historical database data, dynamic expert knowledge base, ore size distribution statistical analysis values ​​and semi-autogenous mill filling rate analysis values ​​collected by the process control system platform are sent to the working condition characteristic reasoning and identification system to infer dynamic expert rules: including the control state, control sequence, change amplitude and duration of the control vector, which are used as the expert control vector; the real-time process detection data are also stored in the historical database;

[0127] (4) Applying the expert control vector to the actual SAG-grinding-rock crushing loop to obtain the actual output value, and establishing an initial model (dynamic transfer function matrix) describing the dynamic characteristics between the process vectors of the SAG-grinding-rock crushing loop based on the expert control vector and the actual output value;

[0128] (5) According to the set value (expected value) and the dynamic characteristic model of the semi-autogenous grinding-rock crushing circuit, the control vector of the intelligent process control algorithm, i.e., the intelligent control vector, is obtained;

[0129] (6) Obtain the values ​​of various parameters in the intelligent controller according to the expert control vector and the intelligent control vector, and use this intelligent controller as the initial intelligent controller;

[0130] (7) Set the validity period of the dynamic characteristic model of the semi-autogenous grinding-rock crushing circuit. When the error between the predicted output value and the actual output value is greater than a certain threshold, the dynamic characteristic model becomes invalid and it is necessary to re-obtain the dynamic characteristic model and the corresponding intelligent controller;

[0131] (8) When the dynamic characteristic model fails, the parameters of the transfer function matrix of the dynamic characteristic model are solved according to the output vector and actual output value of the intelligent control algorithm in the previous validity period, and used as the model parameters in the next validity period to realize dynamic update of the model. After the dynamic characteristic model of the semi-autogenous grinding-rock crushing circuit is dynamically updated, the parameters of the intelligent process controller are updated according to the set value (expected value) and the expert control vector, so as to realize the dynamic update of the entire semi-autogenous grinding-rock crushing circuit intelligent control method and device.

[0132] The semi-autogenous grinding-rock crushing circuit working condition is mainly to identify the working state of the semi-autogenous grinding machine, and the axial pressure, power and filling rate of the semi-autogenous grinding machine are the main parameters reflecting the working state of the semi-autogenous grinding machine. Therefore, the semi-autogenous grinding-rock crushing circuit working condition can be judged according to its axial pressure, power and filling rate. Among them, the priority of the filling rate is the highest, the priority of the axial pressure is the second, and the priority of the power is the lowest, that is, the filling rate is within a reasonable range, which is the constraint condition that needs to be met first. During normal production, the filling rate is usually within a reasonable range, and the axial pressure and power are more sensitive to changes in the working conditions and need to be adjusted frequently. Therefore, what is provided in the embodiment is that when the filling rate is within a reasonable range, if the axial pressure and power of the semi-autogenous grinding machine are controlled.

[0133] During the production process, it is usually necessary to classify the process or equipment parameter values, namely high-high, high, normal, low, and low-low. According to the real-time values ​​and change rules of power and shaft pressure, the classification is shown in Table 1.

[0134] Table 1 Classification of axial pressure and power of semi-autogenous mill

[0135]

[0136] According to Table 1, the axial pressure and power of the SAG mill are arranged and combined in sequence (AA~MM), and the working state of the SAG mill, that is, the working condition of the SAG mill-rock crushing circuit, can be obtained. For example, when the real-time values ​​of the axial pressure and power are both within the normal range, the working condition is as shown in Table 2.

[0137] Table 2 Classification of semi-autogenous grinding-rock crushing circuit operating conditions

[0138]

[0139] The control vectors of the semi-autogenous grinding circuit are the semi-autogenous grinding feed rate, feed size distribution, water addition, semi-autogenous grinding machine speed, and steel ball addition. The key to the operating condition characteristic reasoning and identification system is to determine the dynamic expert rules according to the operating conditions: including the control state, control sequence, change amplitude, and duration of the control vector.

[0140] Due to differences in operating habits, ore properties, production goals, etc., the control strategies of the semi-autogenous grinding-rock crushing circuits of different concentrators are not exactly the same for the same working conditions. Therefore, it is necessary to study targeted expert control strategies based on specific concentrators. This paper takes a concentrator in Yunnan as an example to study the intelligent control method and device of the semi-autogenous grinding-rock crushing circuit. The concentrator does not have an automatic ball adding machine, and the steel ball adding strategy is 2 times / day (two shifts a day, one time per shift). Therefore, the amount of steel ball added cannot be used as a real-time control quantity. The control vectors in the actual control strategy are the four quantities of semi-autogenous grinding feed, feed block size distribution, water addition, and semi-autogenous grinding machine speed. The primary production goal of the semi-autogenous grinding circuit of the concentrator is to ensure that the processing volume meets the production scheduling requirements under the premise of safe operation of the equipment, and secondly to achieve the minimum power consumption per ton of the semi-autogenous grinding machine. There is no requirement for the particle size and concentration of the semi-autogenous grinding machine discharge product (online detection is not achieved).

[0141] For different working conditions, the priorities and initial values ​​of the above four control vectors are also different. For example, the working condition with working condition identification code EE in Table 2 can be described as the real-time value of the axial pressure of the semi-autogenous mill is within the normal range, and the change trend is increasing. At the same time, the real-time value of the power of the semi-autogenous mill is within the normal range, and the change trend is increasing. This working condition indicates that the weight of the material in the semi-autogenous mill is increasing, resulting in an increase in the axial pressure and power at the same time. In order to stabilize its change trend, priority is given to speeding up the discharge speed of the semi-autogenous mill. The corresponding control strategy is to reduce the concentration of the semi-autogenous mill discharge product that is not required for production (increase the amount of water added) and increase the speed of the semi-autogenous mill (increase the motor frequency of the semi-autogenous mill). Analysis of the working condition again shows that: when the feed rate of the semi-autogenous mill remains stable, the weight of the material in the semi-autogenous mill is increasing, indicating that the grindability of the ore is reduced. At this time, the change trend of the semi-autogenous mill shaft pressure and power can also be stabilized by reducing the feed block size. However, reducing the feed block size requires switching the heavy plate feeder, which has a greater impact on production. Therefore, the priority is lower than increasing the amount of water added and increasing the speed of the semi-autogenous mill. In summary, the dynamic expert rules for the working condition with the working condition identification code EE in Table 2 are: the semi-autogenous mill feed amount remains unchanged, and the entire control process is divided into two time periods. In the first time period, the amount of water added to the semi-autogenous mill entrance is increased and the motor frequency of the semi-autogenous mill is increased; when the control effect does not reach the set value, the second time period is started, and the semi-autogenous mill feed block size is reduced on the basis of increasing the amount of water added to the semi-autogenous mill entrance and increasing the motor frequency of the semi-autogenous mill until the working condition changes, and then the dynamic expert rules after the working condition changes are inferred based on the working condition characteristics inference recognition system. The dynamic expert rules of this working condition are shown in Table 3.

[0142] Table 3 Dynamic expert rules for SAG-rock crushing circuit

[0143]

[0144] By control status K , control sequence L , Change Range U and duration T The expert control vector is obtained by calculation V ;

[0145] The expert control vector V Acting on the actual semi-autogenous grinding-rock crushing circuit to obtain the actual output value Yo , according to the expert control vector V and the actual output value Yo Establish an initial model (dynamic transfer function matrix) to describe the dynamic characteristics between the process vectors of the SAG-rock crushing circuit G ( s );

[0146] Let the actual output value Yo (s ) is equal to the set value (expected value) X ( s ),Depend on Yo ( s )= G ( s ) W ( s ) Solution W ( s ),Right now

[0147] W ( s )= G -1 ( s ) Yo ( s )

[0148] and V ( s ) is also known, and can be based on W ( s )= D ( s ) V ( s ) Solution D ( s ),Right now

[0149] D ( s )= W ( s ) V -1 ( s ) = [ G -1 ( s ) Yo ( s )] V -1 ( s )

[0150] Through the above calculation method, the parameters of the intelligent process controller can be solved, and this intelligent controller is used as the initial intelligent controller;

[0151] Set the validity period of the dynamic characteristic model of the semi-autogenous grinding-rock crushing circuit. When the error between the predicted output value and the actual output value is greater than a certain threshold, the dynamic characteristic model becomes invalid and the dynamic characteristic model and the corresponding intelligent controller need to be re-obtained.

[0152] When the dynamic characteristic model fails, the transfer function matrix of the dynamic characteristic model is solved according to the output vector and actual output value of the intelligent control algorithm in the previous validity period G ( s) as the model parameters in the next validity period to achieve dynamic model update;

[0153] After the dynamic characteristic model of the semi-autogenous grinding-rock crushing circuit is dynamically updated, according to the set value (expected value) X ( s ) and the expert control vector V ( s ) Update the intelligent process controller D ( s ) parameters to achieve dynamic update of the entire semi-autogenous grinding-rock crushing circuit intelligent control method and device.

[0154] It can be seen that the intelligent control solution of the semi-autogenous grinding-rock crushing circuit provided in the embodiment of the present application has the following specific characteristics and advantages:

[0155] 1. In addition to the process control system platform, add an intelligent control system platform;

[0156] 2. A dynamic expert knowledge base has been designed. When the working conditions change significantly or the working conditions outside the expert control rules appear, effective control of the working conditions can be achieved by adding expert knowledge and operation rules online;

[0157] 3. Designed a semi-autogenous grinding-rock crushing circuit intelligent control structure, which integrates expert analysis and decision-making, circuit dynamic modeling, and intelligent process control algorithm dynamic update modules, realizing the combination of expert decision-making and intelligent control;

[0158] 4. The online analysis values ​​of the semi-autogenous mill filling rate and the ore size distribution are integrated into the reasoning and identification of working condition characteristics as important process indicators, which increases the dimension of working condition description and improves the accuracy of reasoning and identification of working condition characteristics;

[0159] 5. Designed the process and algorithm for inference and identification of working condition characteristics, namely, the control state of the control vector K , control sequence L , Change Range U and duration T The expert control vector is obtained by calculation V ;

[0160] 6. A modeling method for the dynamic characteristics model of the semi-autogenous grinding-rock crushing circuit is proposed, that is, the expert control vector V Acting on the actual semi-autogenous grinding-rock crushing circuit to obtain the actual output value Yo , according to the expert control vector V and the actual output value Yo Establish an initial model (dynamic transfer function matrix) to describe the dynamic characteristics between the process vectors of the SAG-rock crushing circuit G ( s );

[0161] 7. A design method for intelligent process control algorithm for this loop is proposed, that is, the actual output value Yo ( s ) is equal to the set value (expected value) X ( s ),Depend on Yo(s) = G(s) W(s) Solution W ( s ),Right now W ( s )= G -1 ( s ) Yo ( s ),and V ( s ) is also known, and can be based on W ( s )= D ( s ) V ( s ) Solution D ( s ),Right now

[0162] D ( s )= W ( s ) V -1 ( s ) = [ G -1 ( s ) Yo ( s )] V -1 ( s )

[0163] 8. A dynamic updating method for the dynamic characteristic model of the semi-autogenous grinding-rock crushing circuit is proposed. That is, the validity period of the dynamic characteristic model of the semi-autogenous grinding-rock crushing circuit is set. When the error between the predicted output value and the actual output value is greater than a certain threshold, the dynamic characteristic model fails and the dynamic characteristic model and the corresponding intelligent controller need to be re-obtained. When the dynamic characteristic model fails, the transfer function matrix of the dynamic characteristic model is solved according to the output vector of the intelligent control algorithm and the actual output value in the previous validity period. G ( s ) as the model parameters in the next validity period to achieve dynamic model update;

[0164] 9. A dynamic update method for intelligent process control algorithm is proposed, that is, after the dynamic characteristic model of the semi-autogenous grinding-rock crushing circuit is dynamically updated, the set value (expected value) is calculated. X( s ) and the expert control vector V ( s ) Update the intelligent process controller D ( s ) parameters to achieve dynamic update of the entire SAG-rock crushing circuit intelligent control method and device;

[0165] 10. The axial pressure change rate and power change rate of the semi-autogenous mill are used as the set value (expected value) and feedback value, together with the axial pressure, power and filling rate as the control vector, which increases the dimension of the control vector and serves as the basis for reasoning and identifying the operating condition characteristics, thereby increasing the accuracy of operating condition identification.

[0166] Corresponding to the above-mentioned grinding process intelligent control method, the embodiment of the present application further provides a grinding process intelligent control system for intelligently controlling a grinding process loop, wherein the grinding process loop refers to a loop composed of at least a grinding machine and a crusher, and the system includes:

[0167] A data acquisition module is used to acquire real-time detection data of the grinding process circuit, online detection data of grinding sound and frequency, and online detection data of ore block size. The real-time detection data includes operating status parameters and equipment parameters of the grinding process circuit;

[0168] The working condition identification and expert rule determination module is used to perform real-time working condition characteristic reasoning and identification according to the reasoning logic of the pre-constructed dynamic expert knowledge base based on the real-time detection data, the historical data obtained from the pre-established historical database, the online analysis value of the mill filling rate obtained by analyzing the online detection data of the grinding sound and grinding frequency, and the online analysis value of the ore fragmentation distribution obtained by analyzing the online detection data of the ore fragmentation, to obtain the current working condition characteristics, and determine the corresponding dynamic expert rules according to the current working condition characteristics;

[0169] A model prediction module is used to obtain an output vector of a dynamic characteristic model of a grinding process loop based on the dynamic expert rules, the real-time detection data, the online analysis value of the mill filling rate, and the online analysis value of the ore size distribution as inputs of an intelligent process control algorithm;

[0170] A control execution module is used to perform closed-loop control on the grinding process loop according to the output vector.

[0171] An embodiment of the present application further provides a storage medium, in which a computer program is stored, wherein the computer program is configured to execute the steps of any of the above method embodiments when running.

[0172] Optionally, in this embodiment, the storage medium may be configured to store a computer program for performing the following steps:

[0173] Real-time acquisition of real-time detection data of the grinding process circuit, online detection data of grinding sound and frequency, and online detection data of ore block size. The real-time detection data includes operating status parameters and equipment parameters of the grinding process circuit;

[0174] Based on the real-time detection data, the historical data obtained from the pre-established historical database, the online analysis value of the mill filling rate obtained by analyzing the grinding sound and grinding frequency online detection data, and the online analysis value of the ore fragmentation distribution obtained by analyzing the ore fragmentation online detection data, according to the reasoning logic of the pre-established dynamic expert knowledge base, real-time working condition characteristic reasoning and identification are performed to obtain the current working condition characteristic, and the corresponding dynamic expert rules are determined according to the current working condition characteristic;

[0175] Based on the dynamic expert rules, the real-time detection data, the online analysis value of the mill filling rate and the online analysis value of the ore size distribution as inputs of the intelligent process control algorithm, an output vector of the dynamic characteristic model of the grinding process loop is obtained;

[0176] The grinding process loop is closed-loop controlled according to the output vector.

[0177] Optionally, in this embodiment, the above-mentioned storage medium may include but is not limited to: a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk, and other media that can store computer programs.

[0178] An embodiment of the present application further provides an electronic device, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0179] Optionally, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0180] Optionally, in this embodiment, the processor may be configured to perform the following steps through a computer program:

[0181] Real-time acquisition of real-time detection data of the grinding process circuit, online detection data of grinding sound and frequency, and online detection data of ore block size. The real-time detection data includes operating status parameters and equipment parameters of the grinding process circuit;

[0182] Based on the real-time detection data, the historical data obtained from the pre-established historical database, the online analysis value of the mill filling rate obtained by analyzing the grinding sound and grinding frequency online detection data, and the online analysis value of the ore fragmentation distribution obtained by analyzing the ore fragmentation online detection data, according to the reasoning logic of the pre-established dynamic expert knowledge base, real-time working condition characteristic reasoning and identification are performed to obtain the current working condition characteristic, and the corresponding dynamic expert rules are determined according to the current working condition characteristic;

[0183] Based on the dynamic expert rules, the real-time detection data, the online analysis value of the mill filling rate and the online analysis value of the ore size distribution as inputs of the intelligent process control algorithm, an output vector of the dynamic characteristic model of the grinding process loop is obtained;

[0184] The grinding process loop is closed-loop controlled according to the output vector.

[0185] Optionally, the specific examples in this embodiment may refer to the examples described in the above embodiments and optional implementation modes, and this embodiment will not be described in detail here.

[0186] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0187] In the above embodiments of the present application, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0188] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0189] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0190] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0191] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or optical disk, etc., which can store program code.

[0192] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A grinding process intelligent control method, characterized in that: For intelligently controlling a grinding process loop, wherein the grinding process loop refers to a loop consisting of at least a grinding machine and a crusher, the method comprising: Real-time acquisition of real-time detection data of the grinding process circuit, online detection data of grinding sound and frequency, and online detection data of ore block size. The real-time detection data includes operating status parameters and equipment parameters of the grinding process circuit; Based on the real-time detection data, the historical data obtained from the pre-established historical database, the online analysis value of the mill filling rate obtained by analyzing the grinding sound and grinding frequency online detection data, and the online analysis value of the ore fragmentation distribution obtained by analyzing the ore fragmentation online detection data, according to the reasoning logic of the pre-established dynamic expert knowledge base, real-time working condition characteristic reasoning and identification are performed to obtain the current working condition characteristic, and the corresponding dynamic expert rules are determined according to the current working condition characteristic; Based on the dynamic expert rules, the real-time detection data, the online analysis value of the mill filling rate and the online analysis value of the ore size distribution as inputs of the intelligent process control algorithm, an output vector of the dynamic characteristic model of the grinding process loop is obtained; Performing closed-loop control on the grinding process circuit according to the output vector; The method further comprises: establishing an initial model of the dynamic characteristic model, wherein an expert control vector of the dynamic expert rule is applied to the grinding process loop to obtain an actual output value, and a dynamic transfer function matrix describing the dynamic characteristics of the grinding process loop is established according to the expert control vector and the actual output value to obtain an initial model; The dynamic characteristic model is a model for describing the relationship between the intelligent control vector and the controlled vector, wherein the intelligent control vector is solved according to the set value and the dynamic characteristic model, and the parameter value of the intelligent controller is solved according to the expert control vector and the intelligent control vector; wherein the intelligent control vector includes at least one of the feed rate, the amount of water added, the rotation rate, the feed block size, and the amount of steel balls added, and the controlled vector includes at least one of the power, the axial pressure, and the filling rate.

2. The method according to claim 1, characterized in that: The real-time operating condition characteristic reasoning and identification includes: Based on the real-time detection data, the historical data, the online analysis value of the mill filling rate, and the online analysis value of the ore size distribution, at least one of the axial pressure, power, axial pressure change rate, power change rate, and filling rate is determined as an identification index to perform real-time operating condition characteristic reasoning and identification.

3. The method according to claim 2, characterized in that The step of determining at least one of the shaft pressure, power, shaft pressure change rate, power change rate, and filling rate as the identification index includes: Determining at least one of an axial pressure setting value, a power setting value, an axial pressure change rate setting value, a power change rate setting value, and a filling rate setting value; and, Determine at least one of an actual value of axial pressure, an actual value of power, an actual value of axial pressure change rate, an actual value of power change rate, and an actual value of filling rate according to an output vector of the dynamic characteristic model; The identification index is determined by combining the set value and the actual value.

4. The method according to claim 1, characterized in that Also includes: Based on the dynamic expert rules, the dynamic expert knowledge base is updated in real time, wherein the dynamic expert knowledge base is used to store and manage expert knowledge of grinding process circuit operating condition identification and operating experience, so as to identify the operating condition type according to the expert knowledge and give corresponding expert rules, and the expert rules include at least one of the control state, control sequence, change amplitude and duration of the intelligent control vector of the dynamic characteristic model.

5. The method according to claim 1, characterized in that Determining the corresponding dynamic expert rules according to the current operating condition characteristics includes: Determine the current operating condition characteristic identification result, wherein the identification result includes at least one of a control state, a control sequence, a change amplitude, and a duration of the intelligent control vector to be generated; According to at least one of the control state, control sequence, change amplitude and duration, an expert control vector of the dynamic expert rule is calculated.

6. The method according to claim 5, characterized in that Also includes: The model parameters of the dynamic characteristic model are solved by the following steps: a) Set the validity period of the dynamic characteristic model, linearize the model within the validity period, and determine that the dynamic characteristic model is a linear steady-state system within this period; b) For a linear steady-state system, the dynamic characteristic model is represented based on a dynamic transfer function matrix and an expert control vector: Yo(s)=G(s)V(s), wherein Yo(s) represents the controlled vector, G(s) represents the dynamic transfer function matrix, and V(s) represents the expert control vector.

7. The method according to claim 1, characterized in that The control process of the intelligent process control algorithm is: According to the dynamic transfer function matrix G(s) of the dynamic characteristic model, the intelligent controller D(s) is designed, and the intelligent control vector W(s) is expressed as: W(s) = D(s)V(s), where V(s) represents the expert control vector; The parameters of the intelligent controller are solved by the following steps: W(s) is solved according to Yo(s)=G(s)W(s), where Yo(s) represents the controlled vector; D(s) is solved according to W(s)=D(s)V(s).

8. The method according to any one of claims 1 to 7, characterized in that: Also includes: Dynamically updating the dynamic characteristic model, wherein: By setting the validity period of the dynamic characteristic model, when the error between the predicted output value and the actual output value is greater than a certain threshold, it is confirmed that the dynamic characteristic model is invalid; When the dynamic characteristic model fails, the parameters of the dynamic transfer function matrix of the dynamic characteristic model are solved according to the output vector and actual output value of the intelligent control algorithm in the previous validity period as the model parameters in the next validity period to dynamically update the dynamic characteristic model.

9. The method according to claim 8, characterized in that After dynamically updating the dynamic characteristic model, the method further includes: Update the parameters of the intelligent controller D(s) according to the preset values ​​X(s) and the expert control vector V(s).

10. An intelligent control system for grinding process, characterized in that: Used to execute the intelligent control method of grinding process described in any one of claims 1-9.

11. A storage medium, characterized in that: The storage medium stores a computer program, wherein the computer program is configured to execute the method according to any one of claims 1 to 9 when executed.

12. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to run the computer program to perform the method according to any one of claims 1 to 9.

Images