Method and mill system for optimizing grinding processes

Abstract

The invention relates to a method for optimizing grinding processes in a mill (2), in particular a vertical roller mill, comprising: creating a digital model of the mill (2); continuously recording operational parameters; determining one or more target values for one or more mill parameters, in particular based on the digital model, the operational parameters and / or an operator input; and feedback-controlling the mill (2) based on the determined target value or target values. The invention additionally relates to a mill system (1) for optimizing grinding processes and to the use of a first control unit (17) and a second control unit (18) in a mill system (1).

Description

[0001] Methods and milling systems for optimizing milling processes

[0002] The present invention relates to a method for optimizing grinding processes in a mill, a mill system for optimizing grinding processes and the use of a first control unit and a second control unit in a mill system.

[0003] In conventional grinding processes, one or more control variables are specified externally, for example, by operator input. The grinding process is then controlled by a control unit, such as a programmable logic controller (PLC), based on the specified control variable.

[0004] German patent DE102012024316A1 discloses a method for controlling the moisture content of the feed material. JP4801552B2 discloses a moisture control system for a biomass shredding device by adjusting the heating temperature of the supplied air. US5386945A discloses a control system for a vertical roller mill based on the particle size distribution of the feed material.

[0005] Existing systems are often not flexible enough to efficiently adapt to variable requirements, such as different materials being ground, and / or changing boundary conditions, for example, due to wear of the grinding tools. This can lead to poor product quality, increased energy consumption, and consequently, increased CO2 emissions.

[0006] The object of the present invention is to provide an optimization of grinding processes for variable requirements and boundary conditions.

[0007] The present invention can increase the sustainability of milling processes by minimizing energy consumption and thus CO2 emissions. Furthermore, it ensures high product quality.

[0008] A first aspect of the invention relates to a method for optimizing milling processes in a mill. The method comprises creating a digital model of the mill, acquiring, in particular continuously acquiring, operating parameters, determining one or more setpoints for one or more mill parameters, and controlling the mill based on the determined setpoint(s). The digital model of the mill can represent a digital twin of the mill. Determining the one or more setpoints for the one or more mill parameters can be based on the digital model and the operating parameters. Determining the one or more setpoints for the one or more mill parameters can additionally be based on operator input. The operator input can be provided on, or as input for, the second control unit described below.In one embodiment, the process steps are carried out in the order mentioned above. In alternative embodiments, the process steps can be carried out in other orders, provided this is technically feasible.

[0009] The operational parameters can include process parameters, mill parameters, and / or quality parameters. The process parameters can include or correspond to the mill parameters.

[0010] Mill parameters can also be referred to as manipulated variables. Mill parameters can, in particular, represent controlled variables. The setpoints of the mill parameters can also be referred to as reference variables. The setpoints can, in particular, represent reference variables to which the manipulated variables or mill parameters are to be adjusted. The mill parameters or manipulated variables can represent quantities that can be set by a mill operator.

[0011] Examples of mill parameters or control variables are the clamping pressure / contact pressure of the grinding rollers, the grinding disc speed, the feed quantity, the temperature after the mill (or burner load), the fan speed (or air volume flow), classifier speed, the recirculation damper position, the chimney damper position and / or the water injection.

[0012] Quality parameters can include target values. Target values ​​are, in particular, values ​​that are to be achieved. The target values, or parts thereof, can correspond to the quality parameters. Examples of target values ​​are throughput, mill vibration, specific grindability, the quality of the finished product / ground material, the fineness of the finished product / ground material, and / or the strength of the finished product / ground material. The target values, or parts thereof, can be included in the operational parameters.

[0013] The operating parameters can include disturbances. Disturbances are, in particular, external factors that cannot be directly influenced, or can only be influenced to a limited extent, by the mill's control system. Examples of disturbances include mill wear, the particle size distribution of the feed material, the moisture content of the feed material, the chemical composition of the feed material, and / or the hardness / grindability of the feed material.

[0014] The operational parameters can include manipulated variables, target variables and / or disturbance variables.

[0015] In particular, a target value can be determined for each operating characteristic. When mill parameters (plural) are mentioned below, this explicitly includes the singular form as well. Therefore, it can refer to one mill parameter or to several mill parameters.

[0016] The mill can be a vertical roller mill. The mill can be a ball mill, a high-pressure roller mill (HPGR mill), a roller crusher, or similar. The mill can include a classifier, in particular a rotary classifier. A classifier, in particular a rotary classifier, can be located downstream of the mill. The mill can include a fan. The fan can be designed to influence the air supply to the mill and / or the air circulation within the mill. The term "mill" is also understood to include, in particular, the combination of a mill (grinding device) and a separation device (classifier), as is the case, for example, with a vertical roller mill.

[0017] Continuous data acquisition can be understood as the automatic, recurring acquisition of operating parameters, for example, at 10 Hertz (Hz), every second, every 10 seconds, every minute, every hour, and / or several times a day. This acquisition can occur at regular or irregular intervals.

[0018] Recording the operating parameters may include recording the ambient temperature using a temperature sensor located outside the mill.

[0019] The acquisition of operating parameters can include measuring the material temperature downstream of the mill using a temperature sensor. The acquisition of operating parameters can include measuring the material temperature upstream of the mill or the temperature of the feed material using a temperature sensor. The acquisition of operating parameters can include measuring the material temperature upstream of the separation device, particularly upstream of the classifier. The acquisition of operating parameters can include measuring the material temperature between the mill and the separation device, particularly the classifier.

[0020] Recording operating parameters can include measuring the mill's vibration using a vibration sensor. The vibration sensor can be mounted on the roller bearings, the grinding plate, or the mill housing, particularly at the gearbox input.

[0021] Recording operational parameters can include recording current energy consumption. Recording operational parameters can also include calculating a CO2 footprint based on energy consumption.

[0022] The acquisition of operational parameters can include measuring the feed quantity with an optical sensor, in particular a camera system. Alternatively, the acquisition of operational parameters can include measuring the feed quantity with a belt scale sensor (in particular a dosing belt scale), which is preferably arranged on a material feed conveyor. The belt scale sensor can detect a belt scale signal indicating the feed quantity in tons per hour. The feed quantity can be controlled via the conveyor belt speed. The feed quantity can be monitored via the belt scale sensor or the dosing belt scale.

[0023] The acquisition of operating parameters can include recording the position of a flap at a material feed opening, in particular whether it is open and / or closed. Specifically, the percentage of the flap open or closed can be recorded. The opening angle of the flap can also be recorded. Furthermore, the acquisition of operating parameters can include recording the rotational speed of a rotary valve for material feeding or discharge, particularly in revolutions per minute. The feed rate can be controlled via the rotational speed of the rotary valve. The rotary valve, in particular, prevents the introduction of false air.

[0024] The acquisition of operational parameters can include the acquisition of the particle size or particle size distribution of the feed material using an optical sensor, in particular a camera system. For this purpose, the feed material acquisition device described in WO2021214230A1 can be used in particular.

[0025] Recording operational parameters can include recording throughput.

[0026] Recording operating parameters can include recording mill differential pressure. Mill differential pressure is, in particular, the pressure difference between a signal at the mill's inlet and a signal at its outlet. Mill differential pressure is primarily influenced by the feed rate (feed rate per unit of time).

[0027] Recording operating parameters can include recording the grinding disc speed, particularly in revolutions per minute. Recording operating parameters can also include recording the clamping pressure of the hydraulic cylinders on the grinding rollers, particularly in bar.

[0028] Recording operating parameters can include recording a fan speed, especially in revolutions per minute, and an air supply.

[0029] Recording operating parameters can include recording the position of an air supply damper, in particular whether it is open and / or closed. Specifically, the percentage of the damper open or closed can be recorded. The damper's opening angle can also be recorded.

[0030] The acquisition of operating parameters can include recording the position of a recirculation damper, in particular its open and / or closed position. Specifically, the percentage of the damper open or closed can be recorded. The opening angle of the damper can also be recorded. The recirculation damper determines, in particular, the proportion of air that is recirculated back into the grinding chamber after being extracted from it. The recirculation damper's position can be used to control the proportion of air recirculated in the process and / or the pressure conditions within the process.

[0031] Recording operating parameters can include recording the position of a chimney damper (also known as a flue damper), specifically whether it is open or closed. In particular, the percentage of the damper open or closed can be recorded, as well as its opening angle. The chimney damper primarily determines the discharge of combustion gases to heat the material being ground, and thus the heat input. The damper position can also be used to control the amount of air recirculated during the process and / or the pressure conditions within the process.

[0032] The recirculation damper position, in combination with the chimney damper position, can be used to control the proportion of recirculated air in the process and / or the pressure conditions in the process.

[0033] Recording the operating parameters can include recording the rotational speed of the separation device, in particular the classifier rotational speed, especially in revolutions per minute.

[0034] Recording operational parameters can include measuring the amount of water introduced via water injection, particularly in cubic meters per hour or liters per minute. Water injection can occur during material feeding and / or within the mill.

[0035] Recording operating parameters can include recording the power output of the separation device, in particular the classifier. Recording operating parameters can include recording the power output of the mill's main drive. Recording operating parameters can include recording the power output of the fan. Recording operating parameters can include recording the current of a bucket elevator.

[0036] Recording operational parameters can include recording the quality of the finished milled material.

[0037] Recording operational parameters can include determining the fineness of the finished milled material, for which a sample of the finished milled material is taken and examined, for example by sieving. Determining the fineness can specifically include calculating a Blaine value and / or a particle size distribution.

[0038] Recording the operating parameters can include recording the residual moisture of the finished milled material, for which in particular a sample of the finished milled material is taken, in which the moisture is measured, for example, with a moisture sensor.

[0039] The recording of operational parameters can include determining the final strength of an end product, for which the finished milled material is further processed, for example, into concrete, and the strength, for example, the compressive strength or splitting tensile strength, of the end product / finished product is then determined, particularly after 7 to 28 days, or after 7 and / or 28 days, for example, of the concrete. This recording can include measuring, reading, interpreting, evaluating, and / or determining. Determining final strength can include determining the hardness, particularly after 7 to 28 days, or after 7 and / or 28 days.

[0040] The operational parameters can be several different operational parameters. It is also possible that only one different operational parameter is recorded. In this case, the operational parameters comprise only one different operational parameter, which is recorded multiple times at time intervals.

[0041] The operating parameters can include throughput, energy consumption, oil consumption, gas consumption, grinding disc speed, clamping pressure / contact pressure of the grinding rollers, feed quantity, particle size distribution in the feed material, ratio of components in the feed material, ambient temperature, ground material temperature after mill, classifier speed, fan speed, clamping pressure of the grinding rollers, water quantity of the water injection, flap position, mill vibration, fineness of the finished ground material, final strength of the end product (e.g. compressive strength of concrete made with ground cement), residual moisture in the finished ground material, as well as combinations thereof.

[0042] In particular, three, four, five, six, seven, eight, nine, ten, or more separate conveyor belts can be provided for material feeding. The following features are described using three conveyor belts as an example. The three conveyor belts can feed different materials to the mill, always in the same ratio. The total feed rate can be adjusted via this ratio. For one product (finished milled material), a constant / predefined ratio of the three conveyor belts to each other is selected or set. Different ratios can be selected for different products. For example, for a first product, the feed rates of the three conveyor belts are set to a first ratio (e.g., 1:1:2). For a second product, the feed rates of the three conveyor belts are set to a second ratio (e.g., 2:1:1).The ratios mentioned are merely examples for illustrative purposes. In practice, the majority of the feed quantity is supplied via one conveyor belt, while additives are supplied via one or more (e.g., two) additional conveyor belts. The individual additives each comprise, in particular, 1% to 10%, preferably 3% to 7%, of the total feed quantity. Exemplary ratios of the feed rates of the three conveyor belts are 0.03:0.07:0.9, 0.05:0.02:0.93, or 0.02:0.03:0.95. The product and / or product quality can thus be controlled by the ratio of the conveyor belts to one another. The total feed quantity (sum of all conveyor belts) is set or predetermined, and therefore, the feed quantities of each individual conveyor belt are also determined by the ratio of the individual conveyor belts to one another.The recording of operating parameters includes, in particular, the recording of each individual feed quantity from the three separate conveyor belts, especially by means of metering belt scales and / or belt scale sensors arranged on the conveyor belts. The operating parameters can include the individual feed quantities from the three separate conveyor belts.

[0043] The digital model of the mill can be a linear or nonlinear function, such as a polynomial function, with one input parameter and one output parameter. The digital model can also be a complex model with multiple input and output parameters. Finally, the digital model can represent a digital twin of the mill. A digital twin is a virtual model that approximates the mill and its behavior. The digital twin can take into account the current state of the mill, such as wear, operating temperature, vibrations, etc.

[0044] The digital model or digital twin can be adapted through machine learning, particularly based on the collected operational parameters. The digital model or digital twin of the mill can be self-learning, especially through the use of a neural network.

[0045] The digital twin of the mill can be a so-called "Basic Digital Twin" (digital representation of an object or system with internal storage and data processing), a so-called "Enriched Digital Twin" (digital twin with integration of neighboring data streams), a so-called "Autonomous Control Twin" (autonomous cyber-physical system with human-machine interface for interaction and intervention), a so-called "Enhanced Autonomous Control Twin" (autonomous control twin with data downstream to subordinate systems), or a so-called "Exhaustive Twin" (autonomous, interoperable system that grants users full intervention and integrates both incoming data and sends outgoing data).

[0046] The digital twin can be generalized and thus used across multiple mills. Creating the digital model, especially the digital twin, can initially involve providing a generalized model, particularly a generalized digital twin. This process can then include adapting (training) the digital model, the digital twin, based on operational parameters. This has the advantage that the mill can be operated from the outset without the need for a test or adjustment cycle. During milling operations, the milling process can then be optimized by adapting the digital model. In particular, process disturbances and / or wear patterns can be detected and incorporated into the digital model.By analyzing the operating parameters and continuously adjusting the digital model, the mill can be controlled and regulated to minimize energy consumption and / or maximize throughput. Furthermore, the mill can be operated in a way that reduces wear.

[0047] One or more mill parameters can include the feed quantity of the feed material, the ratio of the components in the feed material, the grinding plate speed, the temperature of the feed material, the temperature (especially of the ground material) after the mill, the water injection, the clamping pressure / contact pressure of the grinding rollers, the classifier speed, the fan speed, the coarse material discharge, the revolutions of a rotary valve (especially for material feed or coarse material discharge), the flap position of an air supply, and combinations thereof.

[0048] Determining the target values ​​can include determining one target value for each mill parameter.

[0049] Determining the target values ​​for one or more mill parameters can be based on the digital model. Using the digital model, especially the digital twin, of the mill, its behavior can be simulated, allowing for the comparison of the effects of numerous mill parameters or combinations thereof. The digital model, particularly the digital twin, enables the optimization of the mill parameter(s).

[0050] The determination of setpoints can be based on the recorded operating parameters. For example, the recorded operating parameters can be fed into the digital model, especially the digital twin. This allows the current operating state of the mill to be taken into account. The setpoints can then be determined based on the digital model and the operating parameters. Additionally, operator input can be incorporated into the digital model. The determination of the setpoints can then be based on the digital model, the operating parameters, and the operator input.

[0051] The specified setpoint can represent a reference input for a control system. Controlling the mill can involve adjusting a controlled variable to the specified setpoint or reference input. Controlling the mill can also involve determining a manipulated variable based on a deviation between the specified setpoint or reference input and the controlled variable. The manipulated variables are communicated to actuators that control the operation of the mill. Examples of actuators include servo motors, valves, or frequency converters.

[0052] The specified setpoints can represent reference variables for a control system. Controlling the mill can involve adjusting several controlled variables to their respective, corresponding setpoints or reference variables. Mill control can be achieved with several separate controllers. Mill control can also be achieved with a multi-variable controller, particularly a MIMO (Multiple Input Multiple Output) controller.

[0053] Mill control can include adjusting the feed rate. Mill control can include adjusting the composition of the feed material. Mill control can include adjusting the grinding plate speed. Mill control can include adjusting the water injection, particularly to the feed material or within the grinding chamber. Mill control can include adjusting the clamping / contact pressure of the grinding rollers. Mill control can include adjusting the classifier speed. Mill control can include adjusting the fan speed. Mill control can include adjusting the damper position of an air supply. Mill control can include adjusting the coarse material discharge between the classifier and the mill. Mill control can include combinations of the above-mentioned adjustment operations.

[0054] A key aspect of the invention is that determining the setpoints and controlling the mill are decoupled from each other, particularly in different control units. This ensures that if the control unit for determining the setpoints fails, the mill control can continue. This increases operational reliability. Furthermore, the method can be operated with an existing local control unit of the mill by specifying the setpoints of the local control unit using an additional control unit.

[0055] By specifying control variables based on the digital model, the downstream control system can continue to operate robustly even if the digital model fails. In the event of a digital model failure, the system can revert to the last transmitted setpoints or control variables. However, if a manipulated variable is specified by the digital model, meaning the control is performed by the digital model and only a controller is present downstream, the mill's control system will fail if the digital model fails.

[0056] In an exemplary embodiment, the feed rate, grinding disc speed, fan speed, clamping / contact pressure of the grinding rollers, classifier speed, air supply damper position, water injection, and the temperature of the material being ground (or, indicatively, the temperature of the airflow) after the mill (or combinations thereof) can be recorded as operating parameters and used as input parameters. Based on this, a setpoint for the temperature of the material being ground after the mill can be determined, and the mill can be controlled based on this setpoint. Additionally, quality parameters of the material being ground or the finished product can be considered when determining the setpoint.

[0057] The setpoint for the temperature of the milled material after the mill is specified, particularly based on the digital model, especially as a reference variable. In this embodiment, the temperature of the milled material represents a controlled variable. Based on the specified setpoint for the temperature of the milled material after the mill, the mill can be controlled in a local control loop, particularly with a local control unit. The temperature after the mill can be controlled via the burner load of a burner located upstream of the mill. The burner load of the burner represents a manipulated variable by which the controlled variable (temperature after the mill) can be influenced, particularly by the local control unit.

[0058] In another exemplary embodiment, the same operating characteristics can be used as input parameters. Based on these input parameters, a setpoint for the mill differential pressure can be determined, and the mill can be controlled based on this setpoint. The setpoint for the mill differential pressure is specified, in particular, based on the digital model, especially as a reference input. In this embodiment, the mill differential pressure represents a controlled variable. Based on the setpoint for the mill differential pressure, the mill can be controlled in a local control loop, particularly with a local control unit. The mill differential pressure can be controlled via the feed rate. The feed rate represents a manipulated variable, which can influence the controlled variable (mill differential pressure), particularly by the local control unit.

[0059] In another exemplary embodiment, the same operating characteristics can be used as input parameters as in the two preceding embodiments. Based on the input parameters, a setpoint for the air volume flow rate can be determined, and the mill can be controlled based on this setpoint after the classifier. The setpoint for the air volume flow rate is specified, in particular, based on the digital model, especially as a reference variable. In this embodiment, the air volume flow rate represents a controlled variable. Based on the setpoint for the air volume flow rate, the mill can be controlled in a local control loop, particularly with a local control unit. The air volume flow rate can be controlled via the fan speed and / or the damper position of the air inlet.The fan speed is a manipulated variable that can influence the controlled variable (mill differential pressure), particularly by the local control unit. The air supply can be controlled via the damper position and / or the fan speed. The damper position and / or the fan speed are manipulated variables that can influence the controlled variable (air volume flow). The air pressure at the mill inlet can be controlled via the position of a recirculation damper or chimney damper. The position of a recirculation damper is a manipulated variable that can influence the controlled variable (air pressure at the mill inlet).

[0060] Several or all of the local control loops described above can also exist in parallel.

[0061] The mill's control system can be deactivated when the mill is started. This allows for quick intervention by the operator.

[0062] Creating the digital model, especially the digital twin, of the mill and determining the setpoints can be done in or with a first control unit. This first control unit can be a programmable logic controller (PLC). Controlling the mill can be done in or with a second control unit. Alternatively, the second control unit can forward the manipulated variables specified by the first control unit to the actuators in the mill. The second control unit can also forward the acquired operating parameters to the first control unit.

[0063] The second control unit can be a second programmable logic controller (PLC).

[0064] The first control unit can be located remotely, in particular more than 10 meters, more than 100 meters, or more than 1000 meters away from the mill. The first control unit can be located in a cloud, in particular in a data center. The digital model can be stored in a cloud. The determination of the setpoints for the mill parameters can be performed in the cloud. The first control unit can communicate with the second control unit. Alternatively, the first control unit can be located in, on, or in the immediate vicinity of the mill. The first control unit can be configured according to the first control unit described in the second aspect of the invention.

[0065] The term "cloud" is also frequently used to refer to a computing cloud, data cloud, or internet cloud. In the cloud, for example, files can be backed up or software applications used without having to be stored or installed on a desktop computer or within a user's own network. A cloud can be understood as interconnected computer networks. The cloud can be located in a public network or a private network.

[0066] The second control unit may be located in, on, or in the immediate vicinity of the mill. The second control unit may be directly connected to the mill or to the mill's actuators. In the event of a failure of the first control unit and / or a break in the connection between the first and second control units, the second control unit can continue to control the mill. The mill can be operated independently of the first control unit using the second control unit. Setpoints for the mill parameters stored in the second control unit can be specified by the first control unit at regular or irregular intervals. If the first control unit fails, the second control unit can continue operation using the stored setpoints. The second control unit may be configured according to the second control unit described in the second aspect of the invention.

[0067] The second control unit can acquire the operational parameters. The second control unit can then transmit the acquired operational parameters to the first control unit. Alternatively or additionally, the first control unit can acquire the operational parameters.

[0068] The second control unit can be connected to or communicate with one or more sensors of the mill. The second control unit can be configured to acquire operating parameters. The second control unit can be configured to transmit the acquired operating parameters to the first control unit. Alternatively or additionally, the first control unit can be connected to or communicate with one or more sensors of the mill. The first control unit can be configured to receive the acquired operating parameters from the second control unit. The first control unit can be configured to acquire the operating parameters.

[0069] The digital model of the mill, in particular its input parameters, can be adapted based on at least some of the recorded operating parameters. This adaptation can be performed continuously, especially at regular intervals.

[0070] By adjusting the digital model or its input parameters, the current state of the mill, such as temperature, wear, vibrations, etc., can be modeled and taken into account. For example, at higher temperatures, it may be necessary to operate the mill with a modified set of mill parameters to ensure optimal milled material quality. Similarly, worn grinding components may require adjusting the grinding plate speed and / or the recirculation of coarse material rejected by the classifier to maintain optimal milled material quality. Furthermore, energy consumption can be optimized by adjusting the digital model.

[0071] One or more mill parameters may be included in the operating parameters. The set of recorded operating parameters may overlap with the set of mill parameters. Each of the one or more mill parameters may be included in the operating parameters. Some of the multiple mill parameters may be included in the operating parameters. The operating parameters may include additional operating parameters that do not correspond to the mill parameters.

[0072] The operational parameters can include one or more process parameters and one or more quality parameters.

[0073] The process parameters can include energy consumption, oil consumption, gas consumption, grinding disc speed, clamping pressure / contact pressure of the grinding rollers, feed quantity, the ratio of components in the feed material, ambient temperature, temperature / ground material temperature after mill (or burner load), classifier speed, fan speed (or air volume flow), recirculation damper position, chimney damper position, water injection, mill wear, particle size distribution of the feed material, moisture content of the feed material, chemical composition of the feed material, hardness / millability of the feed material and / or mill vibration.

[0074] The quality parameters can include parameters of the finished ground material, the throughput, the specific grindability, the fineness of the finished ground material, the final strength of the final product (e.g. compressive strength of concrete made with ground cement) and / or the residual moisture in the finished ground material.

[0075] The acquisition of operational parameters can include the acquisition of process parameters with a first sampling rate and the acquisition of quality parameters with a second sampling rate.

[0076] The first sampling rate can differ from the second sampling rate. The first sampling rate can be higher than the second sampling rate. In particular, process parameters are recorded more frequently than quality parameters.

[0077] The first sampling rate can be 0.01 Hz (Hertz) to 100 Hz, preferably 0.1 to 10 Hz. The second sampling rate can be 1 to 10 times per day, preferably 2 to 5 times per day.

[0078] Process parameters can be used to adapt an artificial neural network. By capturing process parameters at a higher sampling rate than quality parameters, a larger dataset can be generated for the process parameters. This larger dataset facilitates the application and training of the neural network. The artificial neural network can be self-learning and can be continuously adapted during mill operation.

[0079] The artificial neural network can be structured as follows, for example. The network can have an input layer, a first and second hidden layer (intermediate layer), and an output layer. The acquired operational parameters (process and / or quality parameters) can serve as input parameters for the input layer. In particular, all acquired operational parameters are used as input parameters. In particular, all acquired process parameters are used as input parameters.

[0080] The calculated parameters of the output layer can serve as a basis for the operating characteristics (especially the process characteristics) and can, in particular, represent manipulated variables for control with the local control unit. The individual layers of the network can be fully interconnected. In particular, the two hidden layers are fully interconnected. Specifically, the input layer is fully interconnected with the first hidden layer, the first hidden layer is fully interconnected with the second hidden layer, and the second hidden layer is fully interconnected with the output layer. Fully interconnected means, in particular, that all nodes / neurons of a first layer are connected to all nodes / neurons of a second, subsequent layer.

[0081] According to an exemplary embodiment, the artificial neural network can have 2 to 10 nodes / parameters, in particular 5 to 8 nodes / parameters, as input parameters in an input layer and 1 to 5 nodes / parameters, in particular only one node / parameter, as output parameters in an output layer. The artificial neural network can have one, two, or three intermediate layers, each with 5 to 20 nodes, in particular 8 nodes, between the input layer and the output layer. For example, the neural network has an input layer with 8 nodes, two intermediate layers with 8 nodes each, and an output layer with one node.

[0082] Quality parameters can be used to fit polynomials. Because quality parameters are acquired at a lower sampling rate than process parameters, only a smaller dataset can be generated for them. This smaller dataset is often insufficient to train a neural network. Fitting polynomials, however, is possible even with a small dataset.

[0083] The polynomials can be, for example, linear polynomials. The polynomials can be, for example, quadratic polynomials. The polynomials can be, for example, cubic polynomials. The polynomials can be, for example, quartic polynomials. Preferably, quadratic and / or cubic polynomials are used.

[0084] In particular, a polynomial of the same degree can be fitted for each quality parameter. Alternatively, polynomials of different degrees can be used for different quality parameters.

[0085] Creating the digital model, especially the digital twin, of the mill can include determining and / or weighting influencing factors. Determining the target values ​​can also include determining and / or weighting influencing factors. Creating the digital model, especially the digital twin, of the mill and determining the target values ​​can include determining and / or weighting influencing factors.

[0086] An influencing factor can, for example, indicate a change in an output variable in response to a change in an input variable. An influencing factor can, for example, indicate how strongly an input variable influences an output variable. The influencing factors can, for example, indicate how changes in one set of recorded operational parameters affect the remaining recorded operational parameters. The influencing factors can, for example, indicate how changes in process parameters affect quality parameters. The influencing factors can be displayed to a mill operator.

[0087] The influencing factors can be determined by differentiating the digital model, especially the digital twin. The influencing factors can be determined by differentiating the digital model, especially the digital twin, at a current operating point. The influencing factors can be determined using a correlation matrix.

[0088] The procedure may include sorting the influencing factors by magnitude. The procedure may include determining the two, three, or four largest / strongest influencing factors. The procedure may include displaying the two, three, or four largest / strongest influencing factors to a mill operator. The procedure may include displaying suggested changes to one or more mill parameters, particularly based on the identified influencing factors. The procedure may include allowing an operator to change mill parameters, particularly based on the suggested changes. Alternatively, the mill parameters may be changed automatically, particularly based on the suggested changes. The procedure may include verifying the effects of the suggested changes.

[0089] The mill parameters can correspond to the process parameters. The mill parameters can include process parameters.

[0090] Determining target values ​​for one or more mill parameters can involve defining and / or determining optimization criteria. Examples of optimization criteria include minimizing energy consumption, maximizing throughput, minimizing vibrations, maximizing the quality of the finished milled product, minimizing wear, minimizing the CO2 footprint, and / or minimizing overall costs. Determining target values ​​can involve considering a single optimization criterion. Determining target values ​​can involve considering multiple optimization criteria. For example, an operator can select and / or weight optimization criteria through user input. The operator can, for instance, assign individual weights to each optimization criterion. In particular, the operator can specify individual weighting functions for each optimization criterion.

[0091] Determining the target values ​​can include identifying optimization potentials, especially using a gradient descent method.

[0092] The process can additionally include receiving external factors. These external factors can be considered when determining the target values. They can also influence the optimization criteria. Examples of external factors include the current electricity price, the current gas price, the availability of cooling medium, the demand for finished regrind, the current price of finished regrind, planned maintenance times, and / or the feed rate.

[0093] Determining the target values ​​for one or more mill parameters can be based on a collaborative agent system. A first agent in the collaborative agent system can be tasked with optimizing a first operational parameter, particularly a first process parameter or quality parameter. A second agent in the collaborative agent system can be tasked with optimizing a second operational parameter, particularly a second process parameter or quality parameter. The first agent can be tasked with maximizing throughput. The second agent can be tasked with minimizing energy consumption.

[0094] The collaborative agent system allows for the creation of customizable weighting functions, particularly for the optimization criteria, which can then be considered when determining target values. This system can be used to optimize these criteria, especially through simultaneous, parallel optimization. The optimization criteria typically interact with each other. For example, increasing throughput can lead to increased energy consumption. Therefore, maximizing throughput and minimizing energy consumption cannot be achieved simultaneously. The collaborative agent system enables the optimization of individual criteria to be balanced.

[0095] A first agent can be assigned to a first optimization criterion or tasked with optimizing the first optimization criterion. A second agent can be assigned to a second optimization criterion or tasked with optimizing the second optimization criterion. A third agent can be assigned to a third optimization criterion or tasked with optimizing the third optimization criterion. A fourth agent can be assigned to a fourth optimization criterion or tasked with optimizing the fourth

[0096] The optimization criteria are defined. Different weightings can be assigned to the individual agents or optimization criteria. The collaborative agent system and / or adaptable weighting functions enable greater flexibility in optimizing the milling process. In particular, it allows the milling process to be optimized not only according to fixed parameters but also according to economic and ecological constraints. This results in a holistic, adaptive control system for the milling process, which enables continuous improvement and adaptation to the constantly changing demands of production and / or the market.

[0097] The process can further include specifying one or more desired output parameters. These output parameters can be specified by operator input. For example, an operator can request that the current or predicted throughput and / or the current or predicted energy consumption / CO2 footprint be displayed. The process can include displaying one or more output parameters.

[0098] The method can further include varying mill parameters and simulating the mill's behavior in a digital model to calculate simulated output parameters. The method can also include comparing the desired output parameter(s) with the simulated output parameters. Finally, the method can include determining the target values ​​for one or more mill parameters based on this comparison.

[0099] The procedure can further include re-determining the setpoint(s) for one or more mill parameters. This determination of the setpoint(s) for one or more mill parameters can be performed continuously. Mill operation can then be controlled based on the respective current setpoints.

[0100] The process steps of recording the operating parameters, determining the target values ​​and controlling the mill based on the target values ​​can be performed cyclically / periodically.

[0101] The procedure can further include adapting or retraining the digital model, in particular the digital twin, of the mill based on the recorded operating parameters. This adaptation or retraining of the digital model can occur at regular intervals, for example, at least once per minute, at least once per hour, at least twice per hour, at least four times per hour, at least every two hours, at least every 12 hours, at least once per day, at least once per week, at least once every two weeks, and / or at least once per month. The adaptation or retraining of the digital model can also be triggered by operator input.

[0102] The recording of operating parameters and the control of the mill based on the specified target values ​​can be carried out in parallel.

[0103] The digital model can include a predictive maintenance model. The method can include the predictive forecasting or estimation of maintenance intervals, particularly based on the digital model. The method can include the predictive forecasting or estimation of wear conditions of components of the mill system, particularly based on the digital model. The first control unit can be configured to predict, estimate, or recommend maintenance intervals, particularly based on the digital model. The first control unit can be configured to predict or estimate the wear condition of components of the mill system, particularly based on the digital model.

[0104] The method can include adaptive real-time optimization of the digital model, particularly in cases of fluctuations in the quality of the input material. The method can also include taking the quality of the input material into account, especially when determining one or more target values.

[0105] A second aspect of the invention relates to a milling system for optimizing milling processes. The milling system comprises a mill, a first control unit, a second control unit, and one or more sensors. The one or more sensors are configured to acquire one or more operating parameters at regular or irregular intervals. The first control unit is configured to determine a setpoint or setpoints for one or more mill parameters based on operator input and / or the acquired operating parameter(s), in particular by means of a digital twin of the mill. The second control unit can be configured to control one or more mill parameters based on the determined setpoint(s). The digital twin of the mill can be stored in the first control unit. The digital twin of the mill can be calculated or generated by the first control unit.be simulated.

[0106] The mill can be a vertical roller mill. The mill can be a ball mill, a high-pressure roller mill (HPGR mill), a roller crusher, or similar. The mill can have a classifier, in particular a rotary classifier. A classifier, in particular a rotary classifier, can be located downstream of the mill. The mill can have a fan. The fan can be designed to influence the air supply to the mill and / or the air circulation within the mill.

[0107] The first control unit can be a first programmable logic controller (PLC). The first control unit can be designed or configured to store or simulate a digital model of the mill. The first control unit can communicate with the second control unit. The first control unit can be configured to communicate the specified setpoints to the second control unit, particularly at regular and / or predetermined intervals. The first control unit can be designed according to the first control unit described in the first aspect of the invention.

[0108] The second control unit can be identical to the first. The second control unit can be a second programmable logic controller (PLC). The second control unit can be directly connected to the mill or to the mill's actuators. In the event of a failure of the first control unit and / or a connection between the first and second control units being interrupted, the second control unit can continue to control the mill. The mill can be operated independently of the first control unit using the second control unit. Setpoints for the mill parameters stored in the second control unit can be specified by the first control unit at regular or irregular intervals. If the first control unit fails, the second control unit can continue operation using the stored setpoints.The second control unit can be designed according to the second control unit described in the first aspect of the invention.

[0109] Each sensor can be assigned to and / or configured to acquire an operating characteristic. Multiple sensors can be assigned to a single operating characteristic. The acquisition of an operating characteristic can be based on multiple sensors. This ensures both more reliable measurement and redundancy. A single sensor can be configured to acquire different operating characteristics. In particular, multiple operating characteristics can be calculated or estimated from the measured value of a single sensor. Different sensors can acquire the different operating characteristics at different intervals. The sensors can have different sampling rates. The sensors can have uniform sampling rates.

[0110] The sensor(s) may include a vibration sensor, a humidity sensor, a temperature sensor, an optical sensor, in particular a camera system, and / or a scale. The sensor(s) may include a sensor for detecting the current energy consumption of the mill. The sensor(s) may be configured like the sensors described in the first aspect of the invention.

[0111] The first control unit can comprise a digital model, in particular a digital twin, of the mill. The first control unit can be configured to simulate the behavior of the mill, particularly with respect to the acquired operating parameters. The first control unit is specifically designed to communicate control variables for the mill system to the second control unit. The digital model, in particular the digital twin, can be configured like the digital model described in the first aspect of the invention.

[0112] The second control unit can be trained or configured to control components, especially actuators, of the mill, in order to control or regulate one or more mill parameters.

[0113] The second control unit can be configured to adjust the clamping / contact pressure of the mill's grinding rollers. The second control unit can be configured to adjust the grinding plate speed. The second control unit can be configured to adjust the feed rate. The second control unit can be configured to adjust the classifier speed. The second control unit can be configured to adjust a fan speed. The second control unit can be configured to adjust a damper position in an air circuit. The second control unit can be configured to adjust the water injection, particularly to the feed material or within the grinding chamber. The second control unit can be configured to adjust the coarse material discharge between the classifier and the mill.

[0114] One or more sensors can communicate with at least one other control unit. One or more sensors can be electrically connected, in particular via cable or wireless connection, to at least one other control unit. The second control unit can be configured to communicate the acquired operating parameters to the first control unit.

[0115] Alternatively or additionally, one or more sensors can communicate with the first control unit. The one or more sensors can be electrically connected to the first control unit, in particular via a wired or wireless connection.

[0116] The second control unit can be an internal or local control unit, located in, on, or in the immediate vicinity of the mill. The first control unit can be an external control unit, located remotely from the mill and / or provided in the cloud.

[0117] The second control unit must be located within 10 meters of the mill. The first control unit can be located further away, in particular more than 10 meters, more than 100 meters, or more than 1000 meters away from the mill. The first control unit can be located in or stored in a cloud. Alternatively, the first control unit can be located in, on, or in the immediate vicinity of the mill.

[0118] The second control unit can be configured to receive the setpoints determined by the first control unit and, in particular, to store them in a memory within the second control unit. If the first control unit fails and / or the connection between the first and second control units is interrupted, the second control unit can continue to regulate the mill based on the stored setpoints. If the mill were controlled directly by the first control unit, the control would also fail if the first control unit failed.

[0119] A third aspect of the invention relates to the use of a first control unit and a second control unit in a milling system, wherein the first control unit is used to specify at least one setpoint for one or more milling parameters, in particular based on a digital twin of the milling system, and wherein the second control unit is used to control the one or more milling parameters based on the at least one specified setpoint. The digital twin of the milling system can be stored in the first control unit. The digital twin of the milling system can be calculated or simulated by the first control unit.

[0120] The first control unit can be used to specify setpoint values ​​for several mill parameters. The at least one setpoint value can comprise one setpoint value for each mill parameter. The first control unit can be designed and configured like the first control unit described in the first or second aspect of the invention. The second control unit can be designed and configured like the second control unit described in the first or second aspect of the invention.

[0121] The method according to the first aspect of the invention can be carried out with a milling system according to the second aspect of the invention. The milling system according to the second aspect of the invention can be used with process steps of the method according to the first aspect of the invention. The use according to the third aspect of the invention can be carried out with a milling system according to the second aspect of the invention and / or process steps according to the first aspect of the invention.

[0122] As used in the description of the various described embodiments and the attached claims, the singular forms are to be understood as also including the plural forms and vice versa, unless the context clearly indicates otherwise.

[0123] The terms "first," "second," "third," and "fourth" are simply to be understood as designations for a specific element or component and do not necessarily indicate a particular order or arrangement of the components or elements mentioned. For example, the presence of a fourth element / component does not necessarily imply the presence of a first, second, or third element / component, and vice versa.

[0124] Advantageous embodiments of the invention are explained below with reference to the accompanying figures. Fig. 1 shows a schematic representation of a mill system according to one embodiment of the invention.

[0125] Fig. 2 shows a schematic representation of an exemplary mill of the mill system in Fig. 1.

[0126] Fig. 3 shows a flowchart of a method according to an embodiment of the invention.

[0127] Fig. 4 shows a flowchart of a method according to an embodiment of the invention.

[0128] Fig. 5 shows a flowchart of a method according to an embodiment of the invention.

[0129] Fig. 6 shows a layout of a digital model according to an embodiment of the invention.

[0130] Fig. 1 shows a schematic representation of a mill system 1 according to the invention. The mill system comprises a mill 2 in the form of a vertical roller mill. Other mill types are also possible without deviating from the essence of the invention. The vertical roller mill 2 comprises a grinding device 3 and a separation device 4. The grinding device 3 comprises a grinding plate 5 and several grinding rollers 6. The separation device 4 is shown by way of example as a rotary classifier.

[0131] Feed material 8, stored in a silo 7, is conveyed via a conveyor belt 9 and a rotary valve 10 to a material feed opening 11 of the vertical roller mill 2. The rotary valve 10 prevents the introduction of false air. The feed quantity of the feed material 8 can be controlled by a mass flow meter 57, in particular in the form of a belt scale, as described below. The feed material 8, also referred to as the material to be ground, is crushed in the grinding unit 3 and then conveyed upwards to the separation unit 4 by an airflow or process gas flow. The airflow or process gas flow can be controlled by a fan 12. Alternatively or additionally to the embodiment shown in Figures 1 and 2, the fan 12 can be arranged downstream of the mill 2, in particular the separation unit 4, and draws air out of the mill.The airflow or process gas flow is then generated in particular by a suction effect.

[0132] The separating device 4 separates the ground material 8 into fines 13 and coarses 14. The fines 13 are discharged from mill 2 via a fines discharge device 15 by means of the airflow. The coarses 14 rejected by the separating device 4 fall back to the grinding device 3, where they are ground again. Optionally, a coarses discharge device 16 is arranged between the separating device 4 and the grinding device. In the present embodiment, the coarses discharge device 16 is shown by way of example as a screw conveyor. The coarses discharge device 16 can discharge a portion of the coarses 14 rejected by the separating device 4 from the vertical roller mill 2. Optionally, all of the coarses 14 rejected by the separating device 4 can also be discharged from the vertical roller mill 2 via the coarses discharge device 16.

[0133] The mill system 1 has a first control unit 17. The first control unit 17 is located remotely from the vertical roller mill 2. In the illustrated embodiment, the first control unit 17 is located in or constituted by a cloud. Alternatively, the first control unit 17 can be located in or on the mill 2, or in the immediate vicinity of the mill. The first control unit 17 can be a programmable logic controller (PLC). The first control unit 17 is configured, in particular, to determine setpoints for mill parameters based on acquired operating characteristics.

[0134] Milling system 1 additionally includes a second control unit 18. The second control unit 18 is located on or near the vertical roller mill 2. The second control unit 18 is configured to acquire operating parameters of milling system 1. The second control unit 18 is configured to transmit the acquired operating parameters to the first control unit 17. The second control unit 18 is configured to receive the specified setpoints from the first control unit 17. The second control unit 18 is configured to control milling system 1, in particular the vertical roller mill 2, based on the received setpoints. For example, the second control unit can control the rotational speed of the grinding plate 5, the contact pressure of the grinding rollers 6, the rotary valve 10, the fan 12, the rotational speed of the rotary classifier 4, and / or the coarse material discharge device 16 to adjust the mill parameters to the received setpoints.

[0135] The vertical roller mill 2 shown in Fig. 1 is shown enlarged in Fig. 2. The second control unit 18 can be configured to control the fan 12, in particular to adjust the speed of the fan 12 to regulate the air flowing through the air inlet 19. The second control unit 18 can be configured to control the fan located downstream of the vertical roller mill 2 (not shown in Figs. 1 and 2), in particular to adjust the speed of the fan to regulate the air flowing through the fines discharge device 15. The mill system 1, in particular the vertical roller mill 2, has a classifier drive 20, in particular in the form of a motor. The classifier drive 20 is designed to regulate the rotation of the separation device 4, in particular a vane wheel 21 of the rotary classifier. The second control unit 18 is configured to control the classifier drive 20.The mill system 1, in particular the vertical roller mill 2, has a gear input shaft 22 which is connected to a mill drive (not shown), in particular in the form of a motor. The mill drive is configured to adjust the rotation of the grinding plate 5, in particular the grinding plate speed. The second control unit 18 is configured to control the mill drive. The mill system 1, in particular the vertical roller mill 2, has a coarse material discharge drive 23, in particular in the form of a motor. The coarse material discharge drive 23 is configured to adjust the speed of the coarse material discharge device 16. The second control unit 18 is configured to control the coarse material discharge drive 23.

[0136] The fine material 13 or coarse material 14 discharged from mill 2 can be sent to laboratory 24. In laboratory 24, quality parameters of the fine material 13 or coarse material 14, such as their fineness or particle size distribution, can be determined. Laboratory 24 can also determine the quality parameters of a finished product, which may be manufactured from the fine material 13 and / or coarse material 14. For example, the hardness, compressive strength, tensile strength, splitting strength, and / or modulus of elasticity of the finished product can be determined.

[0137] Furthermore, Figures 1 and 2 show exemplary and schematic representations of sensors for recording operating parameters. A first temperature sensor 50 is arranged, in particular, outside the mill 2 and records the ambient temperature. A second temperature sensor 51 can be arranged at the outlet of the mill 2, in particular in the fines discharge device 15, and records the temperature of the milled material after mill 2. A first pressure sensor 52 can be arranged at the outlet of the mill 2, in particular in the fines discharge device 15, and records the pressure at the outlet of the mill 2. A second pressure sensor 53 can be arranged at the air inlet 19 and records the pressure upstream of the mill 2. The mill differential pressure can be determined from the difference between the measured values ​​of the second pressure sensor 53 and the first pressure sensor 52. Further pressure sensors 54 can be arranged in the inlet of the mill 2 and / or in the coarses discharge device 16.

[0138] A vibration sensor 55, in particular in the form of an accelerometer, can be arranged on the mill drive, on the gearbox input shaft 22, inside the gearbox, on the various gearbox stages, on the housing of the mill 2 and / or on the classifier drive 20 in order to measure the vibrations of the respective components.

[0139] One or more microphones 56 can be arranged outside the mill 2 and detect the operating noises, in particular by measuring the sound level.

[0140] Mass flow meters 57 can be arranged on the conveyor belt 8, in the fines discharge device 15 and / or the coarses discharge device 16, and measure the feed quantity of the feed material 8, the discharge quantity of the fines 13, or the discharge quantity of the coarses 14. The mass flow meter 57 on the conveyor belt 8 is, in particular, a belt scale. The mass flow meters 57 on the fines discharge device 15 and / or the coarses discharge device 16 are, in particular, volumetric flow meters.

[0141] A level sensor 58 can be arranged in or on the silo 7 and detect the fill level of the feed material 8. The level sensor 58 is, in particular, a radar or ultrasonic sensor. Alternatively or additionally, the level sensor 58 can be a load cell on the silo.

[0142] Speed ​​sensors 59 and / or frequency converters can be arranged on the mill drive, the classifier drive 20, and / or the fan 12 and detect the rotational speeds of the respective components. The rotational speed of the mill drive is, in particular, equal to or proportional to the grinding disc rotational speed. The rotational speed of the classifier drive 20 is, in particular, equal to or proportional to the classifier rotational speed. Furthermore, power meters (not shown) can be arranged on the mill drive, the classifier drive 20, and / or the fan 12 and detect the power of the respective component and / or are calculated from frequency converter signals. The classifier rotational speed, in particular, is determined via a frequency converter.

[0143] One or more moisture sensors 60 can be arranged on the conveyor belt 9 and detect the moisture content of the feed material 8. Moisture sensors 60 can also be arranged in the laboratory 24 to determine the moisture content of the fine material 13, the coarse material 14 and / or the finished product.

[0144] Laboratory 24 can be equipped with a particle size analyzer 61, which determines the product fineness and / or the particle size distribution of the fines 13, the coarses 14, and / or the finished product. Alternatively or additionally to the particle size analyzer 61, a sieve system 62 can also be used to determine the product fineness and / or the particle size distribution.

[0145] A torque sensor 63 can be arranged on the mill drive and detects the torque transmitted from the mill drive to the mill, in particular the grinding plate 5. A torque sensor 63 can also be arranged on the classifier drive 20 and detects the torque transmitted from the classifier drive 20 to the separation device 4.

[0146] An imaging system 64 can be arranged in the area of ​​the conveyor belt 9 and can, in particular, detect the particle size, especially the particle size distribution, of the feed material 8. The imaging system 64 can detect the composition of the feed material 8.

[0147] The described sensors 50 to 64 can be connected to the first control unit 17 and / or the second control unit 18. Preferably, the sensors 60, 61, 62 arranged in the laboratory 24 are connected only to the first control unit 17 and serve, in particular, as input parameters for the digital model of the mill 2 (the digital twin), but are not directly considered in the control of the mill 2 by the second control unit 18. Alternatively, the measured values ​​of the sensors 60, 61, 62 are recorded manually and / or manually entered into the mill system 1, in particular into the first control unit 17.

[0148] Fig. 3 shows a flowchart of a method according to an embodiment of the invention. In a first step 110, a digital model of the mill 2 is created.

[0149] In a second step, 120 operating parameters of mill 2 or mill system 1 are recorded.

[0150] In a third step, at least some of the operating parameters are fed into the digital model as input parameters. The milling behavior of mill 2 is simulated in the digital model. By using the operating parameters, the current state of mill 2 can be taken into account in the simulation. Additionally, operator input is possible. This input allows the operator to specify which parameters should be optimized. For example, an operator can select and / or weight optimization criteria. The operator can, for instance, assign individual weights to each optimization criterion. In particular, the operator can define individual weighting functions for the individual optimization criteria. For example, an operator can specify that the throughput should be maximized, but a defined product quality should not fall below a certain level.Based on the simulation, target values ​​for mill parameters are determined. The operations of step 130 take place in the first control unit or are performed by the first control unit 17.

[0151] In a fourth step, the determined setpoints are transmitted to the second control unit, 18. The second control unit regulates the mill, 2, based on these setpoints. The setpoints serve as parameters for the control variables.

[0152] Fig. 4 shows a flowchart of a method according to an alternative embodiment of the invention.

[0153] In contrast to the embodiment shown in Fig. 3, operating parameters of mill 2 are acquired in a preceding step. Step 100 can be designed identically to step 120. The operating parameters acquired in step 100 are then used in step 110 to create the digital model of mill 2. In particular, the digital model of the mill can be trained based on the acquired operating parameters.

[0154] Fig. 5 shows a flowchart of a method according to a further embodiment of the invention. The process shown in Fig. 5 is an extension of the processes shown in Figs. 3 or 4. The preceding step 100 is therefore shown as optional (dashed line) in Fig. 5. Compared to the embodiments shown in Figs. 3 and 4, the method includes an additional optional fifth step 150. The fifth step 150 is shown by way of example after the second step 120, but it can also be performed after the third step 130 or the fourth step 140. The designations of the individual steps 110, 120, 130, 140, 150 therefore do not necessarily indicate a chronological sequence, but primarily serve to distinguish the individual steps 110, 120, 130, 140, 150 from one another.

[0155] In step 150, the digital model is adapted or retrained based on the operating parameters recorded in step 120. This retraining allows the digital model to be better adapted to the actual mill behavior, thereby improving the optimization of the milling process.

[0156] As shown in Fig. 5, the process is primarily cyclical. As shown in Fig. 5, after the fourth step 140, the second step 120, namely the acquisition of the operating parameters, is repeated. The newly acquired operating parameters are then used again in the third step 130 to determine the setpoints. Subsequently, in the fourth step 140, the control of mill 2 is performed again. The fifth step 150, namely the adjustment of the digital model of mill 2, can be inserted optionally. For example, the adjustment of the digital model of mill 2 in the fifth step 150 can occur only every 10 cycles or only every hour. These values ​​are merely examples. Furthermore, steps 120, 130, and 140 are not necessarily sequential but can also run in parallel.For example, the operating parameters can be continuously recorded (second step 120) while mill 2 is being controlled (fourth step 140). The simulation of the mill's behavior in the digital model can also be performed in parallel. In particular, it is possible that new setpoints for the mill parameters are determined (third step 130) and transferred to the second control unit 18 only when the mill's behavior deviates from its previous state, when the operating parameters change, and / or when an operator input is received.

[0157] Fig. 6 shows a layout of a digital model according to an embodiment of the invention. In the illustrated embodiment, the digital model is an artificial neural network. The artificial neural network 200 has an input layer 210, a first hidden layer 220, a second hidden layer 230, and an output layer 240. In this exemplary network 200, the input layer 210 has five nodes 211-215. Each node 211-215 corresponds to an input parameter.

[0158] For example, the first node 211 can be assigned the detected feed quantity. The second node 212 can be assigned a determined value for the particle size of the feed material 8. The third node 213 can be assigned the detected throughput of mill 2. The fourth node 214 can be assigned the detected fan speed. The fifth node 215 can be assigned the detected speed of the grinding plate 5. The first hidden layer 220 comprises eight nodes 221-228. Each of the eight nodes 221-228 of the first hidden layer 220 is connected (fully meshed) to each of the nodes 211-215 of the input layer. The second hidden layer 230 comprises eight nodes 231-238. Each of the eight nodes 231-238 of the second hidden layer 230 is connected (fully meshed) to each of the eight nodes 221-228 of the first hidden layer 220.

[0159] The output layer 240 includes one node 241. Node 241 of output layer 240 is fully interconnected with all eight nodes 231-238 of the second hidden layer 230. Output layer 240 can also include multiple nodes (not shown) which are connected to all nodes 231-238 of the second hidden layer.

[0160] All connections between the individual nodes 211-241 are assigned weights or weighting functions (no reference symbol in Fig. 6). The weights or weighting functions are determined by training the neural network with a training dataset.

[0161] The value of node 241 can, for example, correspond to the setpoint for the temperature after the mill, the setpoint for the mill differential pressure, or the setpoint for the airflow after the classifier. It is also possible that output layer 240 has multiple nodes and provides setpoints for several mill parameters, particularly process parameters.

[0162] The value output by output layer 240 at node 241 is used as a reference value for the control of mill 2 by the second (local) control unit 18. Alternatively or additionally, the value output at node 241 can be sent to the operator.

[0163] Alternatively, the value of node 241 can correspond to a target variable or quality parameter, such as throughput, mill vibration, specific grindability, the quality of the finished product / ground material, the fineness of the finished product / ground material, or the strength of the finished product / ground material. Nodes 211-215 of the input layer 210 can correspond to mill parameters, process parameters, or control variables. For example, nodes 211-215 of the input layer 210 can correspond to the clamping pressure / contact pressure of the grinding rollers, the grinding disc speed, the feed rate, the temperature after the mill (or burner load), the fan speed (or air volume flow), the classifier speed, the recirculation damper position, the chimney damper position, and / or the water injection. The input values ​​of nodes 211-215 can be varied, the mill behavior can be simulated by the neural network, and thus the effects of the input values ​​of nodes 211-215 (e.g.,The manipulated variables are estimated against the value of node 241 of output layer 240 (or the multiple nodes of output layer 240), i.e., for example, a target variable. If a desired value of node 241 (e.g., a specific product quality) is reached in the simulation, the corresponding input values ​​of nodes 211-215 can be specified as setpoints or reference values ​​of the second control unit 18 and / or output to the operator. Input layer 210 can also have multiple nodes. In particular, each mill parameter or manipulated variable can be assigned a node of input layer 210.

Claims

Claims 1. Method for optimizing grinding processes in a mill (2), in particular a vertical roller mill, comprising: - Creating a digital model of the mill (2), wherein the digital model of the mill (2) represents a digital twin of the mill (2), - continuous recording of operational parameters, - Determining one or more setpoints for one or more mill parameters, based on the digital model and the operating characteristics, and in particular additionally on operator input, and - Rules of the mill (2) based on the specified setpoint(s).

2. Method according to claim 1, wherein the creation of the digital model of the mill (2) and the determination of the setpoints are carried out in or with a first control unit (17) and the control of the mill is carried out in or with a second control unit (18).

3. Method according to claim 1 or 2, wherein the digital model of the mill (2), in particular input parameters of the digital model, is adapted based on at least a part of the recorded operating characteristics.

4. Method according to one of the preceding claims, wherein one or more mill parameters are included from the operating parameters, in particular from process parameters.

5. Method according to one of the preceding claims, wherein the acquisition of operational characteristics comprises the acquisition of process characteristics with a first sampling rate and the acquisition of quality characteristics with a second sampling rate, wherein the second sampling rate is different from the first sampling rate, and wherein in particular the process characteristics are used to fit an artificial neural network and / or the quality characteristics are used to fit polynomials.

6. Method according to one of the preceding claims, wherein creating the digital model of the mill (2) and / or determining the one or more setpoints includes determining and / or weighting influencing factors, wherein the influencing factors are determined in particular by differentiating the digital model at a current operating point.

7. Method according to one of the preceding claims, wherein the determination of one or more setpoints for one or more mill parameters is based on a collaborative agent system, wherein a first agent of the collaborative agent system is tasked with optimizing a first operating characteristic, in particular maximizing throughput, and wherein a second agent of the collaborative agent system is tasked with optimizing a second operating characteristic, in particular minimizing energy consumption.

8. A method according to any of the preceding claims, wherein the method further comprises: - Specification of one or more desired output parameters, in particular through operator input, - Variation of mill parameters and simulation of mill behavior in the digital model of the mill to calculate simulated output parameters, - Matching the desired output parameter(s) with the simulated output parameters, and - Determining one or more target values ​​for one or more mill parameters based on adjustment.

9. Mill system (1) for optimizing milling processes, comprising a mill (2), in particular a vertical roller mill, a first control unit (17), a second control unit (18), and one or more sensors (50-64), wherein the one or more sensors (50-64) are configured to acquire one or more operating parameters at intervals, wherein the first control unit (17) is configured to determine one or more setpoints for one or more mill parameters by means of a digital twin of the mill (2) based on operator input and / or the acquired operating parameters, and wherein the second control unit (18) is configured to control the one or more mill parameters based on the one or more determined setpoints.

10. Mill system according to claim 9, wherein the first control unit (17) comprises a digital model of the mill (2) and / or is configured to simulate the behavior of the mill (2), in particular based on the acquired operating parameters.

11. Mill system according to claim 9 or 10, wherein the second control unit (18) is configured to control components of the mill (2) in order to control or regulate, in particular, one or more mill parameters of the mill (2).

12. Mill system according to one of claims 9 to 11, wherein the one or more sensors (50-64) are in communication with at least the second control unit (18), and the second control unit (18) is configured to communicate the detected operating parameters to the first control unit (17) or the sensors (50-64) are additionally in communication with the first control unit (17).

13. Mill system according to one of claims 9 to 12, wherein the second control unit (18) is an internal control unit, which is in particular arranged in or on the mill (2), and the first control unit (17) is an external control unit, which is arranged remotely from the mill (2) and is in particular provided in a cloud.

14. Mill system according to one of claims 9 to 13, wherein the second control unit (18) is configured to receive the one or more setpoints determined by the first control unit (17) and to store them in a memory of the second control unit (18).

15. Use of a first control unit (17) and a second control unit (18) in a mill system (1), wherein the first control unit (17) is used to specify at least one setpoint for one or more mill parameters based on a digital twin of the mill system (1), and wherein the second control unit (18) is used to control the one or more mill parameters based on the at least one specified setpoint.

16. Method according to any one of claims 1 to 9, wherein the creation of the digital model of the mill (2) and the determination of the setpoints are carried out with a control unit (17), wherein the control unit (17) is located more than 10 meters away from the mill and / or in a cloud.

17. Method according to any one of claims 1 to 9 or 16, wherein the creation of the digital model comprises providing a generalized model and adapting the digital model based on the operational parameters.

Images