Method for controlling a particle-forming fluidization process taking place in a fluidization apparatus

Abstract

The invention relates to a method for controlling a particle-forming fluidization process taking place in a fluidization apparatus with regard to at least one product property (w) of a process item.

Description

[0001] - Method for controlling a particle-forming fluidization process taking place in a fluidization apparatus

[0002] The invention relates to a method for controlling a particle-forming fluidization process taking place in a fluidization apparatus with regard to at least one product property of a process material.

[0003] Typically, according to a state-of-the-art technology not documented in print, particle-forming fluidization processes are operated using process parameter sets, so-called "recipes." The process parameter sets contain process parameters that lead to the desired product properties when a predetermined sequence of process parameters is maintained, so that the fluidization process always follows the same chronological sequence.

[0004] Furthermore, typical product properties, such as the absolute humidity of the particles, can be adjusted by controlling a process parameter, such as the drying gas temperature or the volume flow of the drying gas. A corresponding method for treating particulate process material in a fluidization apparatus, as well as the associated fluidization apparatus, is disclosed in German patent application DE 102020 208 204 B3.

[0005] Despite an approximation to the desired target value with regard to the product property to be controlled, the known processes still show deviations in this regard, which negatively influence the product quality of the process material.

[0006] The object of the invention is therefore to develop an improved method for controlling the particle-forming fluidization process taking place in the fluidization apparatus, which further optimizes the product properties of the process material with regard to the target values.

[0007] This object is achieved in a method of the type mentioned at the outset in that, in a process cycle, a plurality of process parameters of the fluidization process are determined at a first point in time, which are transmitted as actual process parameter values ​​to a control device having a control functionality, wherein in the control device, based on a process model stored for the at least one product property with the actual process parameter values, a process model product property value is calculated for a second point in time following the first point in time, and wherein in the control device, a plurality of optimization value sets are formed from a plurality of provided process parameter optimization values, using which sets of optimization values ​​are calculated at a third point in time by means of an optimization model,wherein each of the optimization preview values ​​is subjected to a correction value to form a corrected optimization preview value, and wherein an optimization difference value is calculated at the third time from a comparison between each of the corrected optimization preview values ​​and a preview setpoint for the at least one product property determined at the third time from a setpoint function stored in the control device, and wherein the process parameter optimization values ​​of the optimization variable set associated with the smallest amount of the optimization difference value are then output as a reference variable for the second time following the first time. Advantageously, the method for controlling a particle-forming fluidization process taking place in a fluidization apparatus without user specifications,Suitable control variables for achieving at least one product property of the process material to be controlled. This significantly increases the product quality of the process material while ensuring high reproducibility of the results.

[0008] In this regard, a plurality of process cycles run consecutively, with the second point in time of each process cycle constituting the first point in time of the subsequent process cycle. Thus, for example, the values ​​of the second point in time in the first process cycle expediently constitute the values ​​of the first point in time of the second process cycle.

[0009] The process parameters are conveniently determined by measurement or by simulation. This has the advantage that the process parameters can be made available in different ways, which can lead to savings in measurement technology, for example, when simulating the process parameters.

[0010] According to a further development of the method, the process parameters are measured as inline measurements and / or at-line measurements and / or online measurements, with the process parameters preferably being measured at a process parameter sampling frequency. The method for controlling the particle-forming fluidization process taking place in the fluidization apparatus with respect to at least one product property of the process material thus always has the current process parameters available for control purposes.

[0011] According to a further refinement of the method, the actual process parameter values ​​determined at the first time form a process parameter value set. This facilitates the transfer of the actual process parameter values ​​to the control device. In this regard, each of the optimization value sets is formed from the plurality of process parameter optimization values ​​corresponding to the plurality of actual process parameter values ​​of the process parameter value set, with at least one process parameter optimization value in the optimization value set substituting a corresponding actual process parameter value.

[0012] Furthermore, each of the process parameter optimization values ​​can assume any desired optimization value, with the optimization value preferably being selectable from a plurality of predefined optimization values. This limits the number of optimization value sets to a number resulting from any combination of process parameters and the plurality of predefined optimization values ​​for the respective process parameter. With the number of predefined optimization values ​​expediently being the same for each process parameter optimization value, the maximum number of optimization value sets results in "number of process parameter optimization values ​​multiplied by the plurality of predefined optimization values."In principle, the number of predefined optimization values ​​is arbitrary, with the number of optimization sets and thus also the number of optimization values ​​being determined by the computing time required for the optimization model in the control system. The more powerful the control system, the more optimization values ​​and process parameter optimization values ​​can be used. Furthermore, a correspondingly high number of optimization values ​​and process parameter optimization values ​​also affects the control accuracy of the fluidization process.

[0013] Advantageously, the specified optimization values ​​are based on the respective actual values ​​of the process parameters. The optimization values ​​represent the largest possible value range within the technological limits of the fluidization process. Optimization values ​​that lead to instabilities in the fluidization process are no longer within the technological limits of the fluidization process. The technological and product-specific limits of the fluidization process with regard to the respective process parameters are usually determined in preliminary tests associated with the fluidization process.

[0014] According to an additional development of the method, there is a first time period containing at least one journal between the first point in time and the second point in time, and a second time period containing at least one journal between the second point in time and the third point in time. In this regard, the first time period and the second time period have a different number of journals, with the first time period expediently containing a single journal. Preliminary tests in this regard have shown that the first time period advantageously has one journal and the second time period 19 time steps. This means that there are 20 time steps between the first and the third point in time. The time period between the first and the third point in time can, however, also be directed further into the future and have a number of, for example, 30, 40, 50 or more journals. The second time period must be set up accordingly.For practical purposes, the number of time steps in the first and second time periods is adjusted to match each other. However, the number of time steps in the first and second time periods remains freely selectable for each individual fluidization process.

[0015] According to a refinement of the method, a setpoint function is stored in the control device for each product property to be controlled. For each product property to be controlled, a setpoint function stored in the control device conveniently maps the desired progression of the respective product property to be controlled over time. For example, with particle size as the selected product property to be controlled, the setpoint function maps particle growth over time.

[0016] According to a further development of the method, the setpoint function for at least one product property to be controlled is formed from experimental data or from a setpoint process model. This adapts the setpoint function to the physical and chemical principles applicable to the particle-forming fluidization process. Furthermore, the method offers the possibility of the setpoint function also representing an arbitrary function specified by an operator and stored in the control device.

[0017] In this regard, the setpoint process model is expediently based on a kinetic model of the at least one product property. A kinetic model refers to the mathematical description of the product property profile of each at least one product property to be controlled in the fluidization process as a function of different process parameters, such as the growth kinetics of the particle size. Preferably, the at least one product property in the process is the particle size and / or the particle moisture content and / or the particle composition. The use of a kinetic model also adapts the setpoint function to the physical-chemical principles applicable to the particle-forming fluidization process.

[0018] According to a further development of the method, the at least one product property to be controlled is recorded as a product property measured at a first point in time and transmitted to the control device as an actual product property value. For this purpose, the actual product property values ​​are smoothed using a mathematical smoothing method, expediently using the Whittaker-Henderson method. Furthermore, the actual product property values ​​form a product property quantity set. This facilitates the transfer of the actual product property values ​​to the control device.

[0019] The correction value is expediently calculated at the first time by subtracting the process model product property value calculated for the at least one product property at the first time from the at least one actual product property value recorded at the first time. By calculating the correction value in this way, the error calculated by the process model in each cycle and accumulating over time is corrected. The correction value at the first time is preferably "0" since no actual product property value is recorded at the first time in the first process cycle. The product properties are preferably recorded as an inline measurement and / or atline measurement and / or online measurement, wherein the product property is expediently recorded at a product property sampling frequency.The current product properties are thus always available for the control process of the particle-forming fluidization process taking place in the fluidization apparatus with regard to at least one product property of the process material.

[0020] In a further preferred development of the method, the process parameter sampling frequency and the product property sampling frequency have the same value. This ensures that the current process parameter actual values ​​and the current product property actual values ​​are provided simultaneously in a process cycle or at a given time.

[0021] Furthermore, if there are two or more product properties to be controlled, a product property size set is created, whereby the product properties to be controlled are prioritized in relation to a higher priority control of one of the product properties. The prioritization of the product properties to be controlled, also referred to as weighting of the product properties to be controlled, is based on the importance of the respective product property in the fluidization process and / or for the product quality and / or the user requirement. If, for example, in a fluidization process the particle moisture to be achieved in the process material is of greater importance than the particle size of the process material, then the particle moisture must be prioritized or weighted accordingly. The prioritization means that an optimization difference value of the prioritized product property is to be achieved with priority compared to a lower priority product property. For this purpose, for example:The optimization difference values ​​of the product properties to be controlled from an optimization variable set are added together, with each optimization difference value being multiplied by a weighting factor according to its prioritization and thus weighted. The sum of the optimization difference values ​​is then divisible by the number of optimization difference values. In one example, the process parameter optimization values ​​of the lowest sum or the lowest average of the optimization variable set associated with the optimization difference values ​​are then output as reference variables.

[0022] According to a further development of the method, the optimization model is based on the process model; in particular, the optimization model corresponds to the process model. An optimization model corresponding to the process model ensures that the multitude of optimization preview values ​​calculated at a third point in time using the optimization model have the same basis as the process model product property value calculated at the first point in time using the process model. Thus, improved control of at least one product property to be controlled can be achieved.

[0023] According to an additional refinement of the method, the process model for calculating the process model product property value is based on a linear or nonlinear process model of the fluidization process to be controlled, with an artificial neural network being used as the nonlinear process model. In this regard, the artificial neural network is designed, in particular, as a multilayer perceptron or as a simple recurrent network, such as an ELMAN network, or as a nonlinear autoregressive exogenous network, such as a NARX network.The artificial neural networks are expediently trained before carrying out the method for controlling a particle-forming fluidization process taking place in a fluidization apparatus with regard to at least one product property of a process material by means of tests carried out in a fluidization apparatus, wherein in the tests carried out, different process parameters are varied with regard to the at least one product property to be controlled.

[0024] As process parameters, preferably one or more process parameters from the group of spray gas pressure and / or spray rate and / or spray quantity and / or particle temperature and / or drying gas temperature at the inlet of the fluidization apparatus and / or the relative humidity of the drying gas at the outlet and / or drying gas volume flow are used.

[0025] The invention is explained in more detail below with reference to the accompanying drawing, which shows

[0026] Figure 1 is a schematic representation of the method for controlling a particle-forming fluidization process taking place in a fluidization apparatus with regard to at least one product property of a process material,

[0027] Figure 2 shows a diagram showing a product property plotted over time in the first process cycle and detail sections A and B,

[0028] Figure 3 is an enlarged view of detail section A,

[0029] Figure 4 is an enlarged view of detail section B, Figure 5 is a diagram showing a product property plotted over time in the second process cycle and detail sections C and D,

[0030] Figure 6 is an enlarged view of detail C and

[0031] Figure 7 is an enlarged view of detail D.

[0032] Unless otherwise stated, the following description refers to all embodiments of a method for controlling a particle-forming fluidization process taking place in a fluidization apparatus with regard to at least one product property w of a process material, as illustrated in the drawing.

[0033] The indices n used in the description refer to any natural number, where n can take different values ​​for different letters. For example, for On, n = 3 and for o''nn = 27.

[0034] In a process cycle z, the sequence of which is explained below using the embodiment shown in Fig. 1, a plurality of process parameters p of the fluidization process are determined at a first time ti. The first time ti is to be understood as the current time t of the fluidization process. The process parameters p can include, among others, the spray gas pressure and / or the spray rate and / or the spray quantity and / or the particle temperature and / or the drying gas temperature at the inlet of the fluidization apparatus and / or the relative humidity of the drying gas at the outlet and / or the drying gas volume flow. The process parameters p are determined by simulation or by measurement. The measurement of the process parameters p is carried out optionally as an inline measurement, atline measurement or online measurement with the aid of an appropriate measuring technique known to those skilled in the art.Conveniently, the process parameters p are sampled with a process parameter sampling frequency f. P measured. The determined process parameters p are transmitted as actual process parameter values ​​p' to a control device 1 having a control functionality. The actual process parameter values ​​p' determined at the first time t1 preferably form a process parameter variable set p''. In an embodiment not shown, some of the process parameters p were simulated, and the other part of the process parameters p were measured.

[0035] In parallel to the measurement and / or simulation of the process parameters p, at least one product property w m at the first time ti with a product property sampling frequency f wmeasured. Conveniently, the measurement is carried out either as an inline measurement, atline measurement, or online measurement. The at least one product property w to be controlled is recorded as an actual product property value w'm measured at a first time ti and transmitted to the control device 1. Conveniently, the actual product property values ​​w'm form a product property size set w''m. The product properties used include, among others, the particle size and / or the particle moisture content and / or the particle composition.

[0036] Particularly preferably, the process parameter sampling frequency f P and the product property sampling frequency f w the same value. As a result, the process parameter actual values ​​p' and the product property actual values ​​w' are located in the control device 1. mat the same time in each case. In the control device 1, the transferred product property actual values ​​w'm are smoothed in a smoothing module 2 assigned to the control device 1 using a mathematical smoothing method. This is expediently done using a mathematical smoothing method such as the Whittaker-Henderson method. The mathematically smoothed product property values ​​w' s subsequently form a product property value set w'' s .

[0037] The control device 1 further comprises a process model module 3 in which, based on a process model stored for the at least one product property w with the recorded actual process parameter values ​​p', which preferably form a process parameter set p'', a process model product property value w'c is calculated for a second time t2 following the first time t1. The process model product property values ​​w' c expediently form a process model product property value set w '' c .

[0038] The process model for calculating the corresponding process model product property value w'c is based on a linear or nonlinear process model of the fluidization process to be controlled, with an artificial neural network being advantageously used as the nonlinear process model. The artificial neural network is preferably designed as a multilayer perceptron, a simple recurrent network, or a nonlinear autoregressive exogenous network.The artificial neural networks are expediently trained before carrying out the method for controlling a particle-forming fluidization process taking place in a fluidization apparatus with respect to at least one product property of a process material by means of tests carried out in a fluidization apparatus, wherein in the tests carried out different process parameters p are varied with respect to the at least one product property w to be controlled.

[0039] Furthermore, the control device 1 has a correction module 4. In the correction module 4, a correction value k is calculated at the first time ti. The correction value k is calculated for each product property w to be controlled by taking at least one actual product property value w'm, preferably the mathematically smoothed actual product property value w' s, the process model product property value w'c calculated for the at least one product property w at the first time ti is subtracted from the at least one product property w. The correction values ​​k can also form a correction value set k''. In the first process cycle zi, the respective correction values ​​k are set to the value "zero" due to the missing process model product property value w'c at the first time ti.

[0040] Furthermore, before the start of the process for controlling a particle-forming fluidization process taking place in a fluidization apparatus, optimization values ​​v are specified with respect to at least one product property w of a process material based on an expected range of validity of the process parameters p. The range of validity here means, for example, a specific range for the particle temperature serving as the process parameter optimization value o, namely, for example, from 75 °C to 100 °C, with the optimization values ​​v dividing the range of validity, in particular equidistantly, for example, into six parts having values, here as values ​​vi = 75 °C, V2 = 80 °C, V3 = 85 °C, v4 = 90 °C, v5 = 95 °C and v s= 100 °C. Generally speaking, each of the process parameter optimization values ​​o can assume any optimization value v, wherein the optimization value v can be selected from a large number of predetermined optimization values ​​v. The large number of optimization values ​​v expediently extends from two optimization values ​​to n optimization values. For this purpose, Fig. 1 shows the optimization values ​​Voi for the process parameter optimization value oi as an example for all optimization values ​​v. In the embodiment shown, the process parameter optimization value o1 has six optimization values ​​Voi.i to Voi,s. Expediently, but not necessarily, all process parameter optimization values ​​o have the same number of optimization values ​​v.

[0041] The control device 1 further comprises an optimization module 5. In the next step, each of the optimization value sets o'' is formed in a combinatorics module 6 assigned to the optimization module 5 from the plurality of process parameter optimization values ​​o corresponding to the plurality of process parameter actual values ​​p' of the process parameter value set p'', wherein in the optimization value set o'' at least one process parameter optimization value o substitutes for a corresponding process parameter actual value p'. In the combinatorics module 6, therefore, for example, with n process parameter optimization values ​​o and six optimization values ​​v o i,i to Voi,s a number of n 6 Optimization size sets o'' are formed. In concrete terms, for example, with three process parameter optimization values ​​o with six optimization values ​​v each, this would result in a number of 3 6= 729 optimization value sets o'' result. The number of optimization values ​​v and the number of process parameter optimization values ​​o depend in particular on the performance of the control device 1. In addition, the optimization module 5 has an optimization model module 7. Using the determined optimization value sets o'', a plurality of optimization preview values ​​x are calculated at a third time ta by means of an optimization model stored in the control device 1, expediently in the optimization model module 7. The plurality of optimization preview values ​​x corresponds to the number of optimization value sets o''. The optimization model is preferably based on the process model; in particular, the optimization model even corresponds to the process model.

[0042] In the next step, each of the optimization preview values ​​x is subjected to a correction value k to form a corrected optimization preview value Xk. In the illustrated embodiment of Fig. 1, the correction value k is subtracted from the optimization preview value x to form the corrected optimization preview value X k subtracted.

[0043] Subsequently, in a comparison module 8 assigned to the control device 1, an optimization difference value A at the third time ta is calculated from a comparison between each of the corrected optimization preview values ​​Xk and a preview setpoint xs for the at least one product property w determined at the third time ta from at least one setpoint function S stored in the control device 1. Preferably, a setpoint function S is stored in the control device 1 for each product property w to be controlled. The setpoint function S for the at least one product property w to be controlled is formed in particular from experimental data or from a setpoint process model. Kinetics of the product property, e.g. growth kinetics of the particle size, can expediently be used to specify the setpoint xs.

[0044] In the embodiment shown in Fig. 1, the optimization difference value A is calculated as the amount of subtraction between the corrected optimization preview value Xk and the target value xs. In the embodiment shown, sets of values ​​are considered.

[0045] The control device 1 further has an evaluation module 9, in which the values ​​of the optimization difference value Δ are evaluated by comparing them with one another. The process parameter optimization values ​​o of the optimization variable set o'' associated with the smallest value of the optimization difference value Δ are then output as a reference variable r for the second time t2 following the first time t1. The respective reference variable r is then applied to the fluidization process and adjusted by means of a further control (P, PI, PID control) of the process parameters p.

[0046] If there are at least two product properties w to be controlled, a product property size set w'' is created, with the product properties w to be controlled being prioritized with regard to a priority control. The product properties w to be controlled are prioritized according to the importance of the respective product property w in the fluidization process. If, for example, the particle size of the process material and the particle moisture content of the process material are to be controlled in the fluidization process with regard to the product property, with the particle moisture to be achieved in the process material being of greater importance than the particle size of the process material, then the particle moisture content must be prioritized or weighted accordingly. For this purpose, the control device 1 has a weighting module 10 in which the prioritization takes place.

[0047] Prioritization means that a prioritized product property w is to be achieved with priority compared to a less prioritized product property w. For this purpose, for example, the amounts of the optimization difference values ​​A of the product properties w to be controlled are added together, whereby each amount of the optimization difference values ​​A is weighted during the addition according to its prioritization, e.g., multiplied by its weighting factor g. The sum of the optimization difference values ​​is then divisible by the number of optimization difference values. In one example, process parameter optimization values ​​o of the lowest sum of the amount of the optimization difference values ​​A of the associated optimization variable set o'' are output as reference variables.

[0048] Between the first time t1 and the second time t2 there is a first time period Ati comprising at least one magazine d, and between the second time t2 and the third time ta there is a second time period Ata comprising at least one magazine d. The first time period Ati and the second time period Ata preferably have a different number of magazines d, with the first time period Ati expediently having a single magazine d.

[0049] In the embodiment shown in Fig. 1, there is a time period Ati comprising one time step d between the first time ti and the second time ta, and a time period Ata comprising 19 time steps d between the first time ta and the third time ta. Thus, there are 20 time steps d between the first time ti and the third time ta. The first and second time periods Ati and Ata can have other value combinations, the aforementioned being the preferred value combination.

[0050] A plurality of process cycles z can run one after the other, whereby the second time ta of the preceding process cycle z forms the first time ti of the subsequent process cycle z+i.

[0051] Fig. 2 shows a diagram with a representation of a product property w plotted over time t in the first process cycle zi and detail sections A and B.

[0052] In the embodiment, control was carried out according to a product property w, with the particle size being chosen as the product property w.

[0053] For the control itself, three process parameters p were used in the illustrated embodiment. The spray gas pressure was used as process parameter p1, the spray rate as process parameter p2, and the particle temperature as process parameter pa. Each process parameter p could only assume two optimization values ​​v.

[0054] Between the first time ti and the second time ta there is a time span Ati comprising one time step d, and between the first time ta and the third time ta there is a time span Ata comprising 19 time steps d. Thus, between the first time ti and the third time ta there are 20 time steps d. Also shown is the setpoint function S for the particle size as a product property w.

[0055] In the combinatorics module 6, with 3 process parameter optimization values ​​o and two optimization values ​​v each, a number of 3 2 Optimization sets of variables o'' are formed. Specifically, nine optimization sets of variables o'' are formed in combinatorics module 6. The nine functions resulting from the optimization model with the respective optimization sets of variables o'' are labeled F1 to Fs in the diagram.

[0056] The optimization preview values ​​xi to xs result from the function values ​​of the nine functions F1 to Fs at the third time point ta.

[0057] In section A of Fig. 3, the course of the measured and the smoothed product property actual value w' m and w's are shown. The smoothed actual product property value w's at time ta is also shown. Furthermore, the actual product property value w'c calculated using the process model is shown at the second time ta.

[0058] Section B of Fig. 4 shows the functions Fs and Fs in an enlarged view and the optimization preview values ​​xs to xs at time ta. At time ta, the two functions Fs and Fs are closest to the setpoint xs of the setpoint function S.

[0059] The correction value k is calculated for the product property w to be controlled by subtracting the process model product property value w'c calculated for the product property w at the first time ti from the mathematically smoothed actual product property value w's at the first time ti. Since the respective correction value k cannot be calculated in the first process cycle zi due to the missing process model product property value w'c at the first time ti, the correction value k is set to the value "zero" in the first process cycle zi.

[0060] In the embodiment shown in Fig. 4, the optimization preview value x5 or x6 is deviated from the target value x s to form the optimization difference values ​​A5 and As. A comparison of the two values ​​of the optimization difference values ​​A5 and As shows that the value of the optimization difference value As is smaller than the value of the optimization difference value A5. Thus, the process parameter optimization values ​​o of the optimization variable set 0''6 associated with the optimization difference values ​​Δ6 are output as the reference variables r for the next time t2.

[0061] Figs. 5 to 7 describe the same as Figs. 2 to 4, but for the second process cycle Z2.

[0062] Fig. 5 shows a schematic representation of a second process cycle Z2 of the exemplary embodiment with a product property w plotted over time t and sections C and D.

[0063] The actual product property value w's and the process model product property value w'c available at the second time t2 of the process cycle z form the actual product property value w's and the process model product property value w'c of the first time ti in the subsequent process cycle z+i. Between the first time t1 and the second time t2 there is a time period Ati comprising one time step d, and between the second time t2 and the third time ta there is a time period Ata comprising 19 time steps d. Thus, as in the first process cycle z1, there are 20 time steps d between the first time ti and the third time ta.

[0064] Also shown is the setpoint function S for the particle size as a product property w.

[0065] In the combinatorics module 6, a number of 3 is determined for the 3 process parameter optimization values ​​o and the two optimization values ​​v. 2Optimization size sets o'' are formed. Specifically, nine optimization size sets o'' are formed in combinatorics module 6. The nine functions resulting from the optimization model with the respective optimization size sets o'' are labeled F1 to F9 in the diagram.

[0066] The optimization preview values ​​x1 to x9 result from the function values ​​of the nine functions F1 to Fs at the third time point ta.

[0067] In section C of Fig. 6, the course of the measured and the smoothed product property actual value w' m and w' s The smoothed product property actual value w' s at time ta is also shown. Furthermore, the actual product property value w'c calculated using the process model is shown at the second time ta.

[0068] Section D of Fig. 7 shows the functions Fs and F? in an enlarged view and the optimization preview values ​​X6 to X7 at time ta. The two functions F6 and F7 are closest to the setpoint xs of the setpoint function S at time ta.

[0069] The correction value k is calculated for the product property w to be controlled by subtracting from the mathematically smoothed product property actual value w' at the first time ti s (t1) the process model product property value w'c(t1) calculated for the product property w at the first time ti is subtracted from the product property w.

[0070] In the embodiment shown in Fig. 7, the correction value k is subtracted from the optimization preview value xs or x?, and then the setpoint xs is subtracted from the corrected optimization preview value x6.k or x7,k to form the optimization difference values ​​Δ6 and Δ7. A comparison of the two absolute values ​​of the optimization difference values ​​Δ6 and Δ7 shows that the absolute value of the optimization difference value Δ7 is smaller than the absolute value of the optimization difference value Δ6. Thus, the process parameter optimization values ​​o of the optimization variable set 0''7 associated with the optimization difference value Δ7 are output as reference variables r for the next time t2.

Claims

Claims 1. A method for controlling a particle-forming fluidization process taking place in a fluidization apparatus with respect to at least one product property (w) of a process material, characterized in that in a process cycle (z) a plurality of process parameters (p) of the fluidization process are determined at a first time point (t1), which are transmitted as actual process parameter values ​​(p') to a control device (1) having a control functionality, wherein in the control device (1) a process model product property value (w'c) for a second time point (t2) following the first time point (t1) is calculated on the basis of a process model stored for the at least one product property (w) with the actual process parameter values ​​(p').and wherein in the control device (1) a plurality of optimization parameter sets (o'') are formed from a plurality of provided process parameter optimization values ​​(o), using which a plurality of optimization preview values ​​(x) are calculated at a third time (t3) by means of an optimization model, wherein each of the optimization preview values ​​(x) is applied with a correction value (k) to form a corrected optimization preview value (xk), and wherein from a comparison between each of the corrected optimization preview values ​​(Xk) and one at the third time (t3) from a in the control device, (1) From the stored setpoint function (S) determined preview setpoint (xs) for at least one product property (w), an optimization difference value (A) is calculated at the third time (t3), and subsequently the process parameter optimization values ​​(o) of the optimization quantity set (o'') belonging to the smallest amount of the optimization difference value (A) are each output as a reference variable (r) for the second time (ta) following the first time (t1).

2. The method according to claim 1, characterized in that a plurality of process cycles (z) run successively, wherein the second time point (ta) of the process cycle (z) corresponds to the first time point (t1) of the following process cycle (z +1 ) forms.

3. Method according to claim 1 or 2, characterized in that process parameters (p) are determined by measurement or by simulation.

4. Method according to claim 3, characterized in that the measurement of the process parameters (p) is carried out as an inline measurement and / or atline measurement and / or online measurement, wherein the process parameters (p) are expediently measured with a process parameter sampling frequency (f P ) are measured.

5. Method according to one of the preceding claims, characterized in that the actual process parameter values ​​(p') determined at the first time (t1) form a process parameter set (p'').

6. Method according to claim 5, characterized in that each of the optimization parameter sets (o'') is derived from the plurality of process parameter actual values ​​(p') of the process parameter- A corresponding number of process parameter optimization values ​​(o) are formed for the set of parameters (p''), wherein at least one process parameter optimization value (o) in the optimization set of parameters (o'') substitutes a corresponding actual process parameter value (p').

7. Method according to one of the preceding claims, characterized in that each of the process parameter optimization values ​​(o) can assume any optimization value (v), wherein the optimization value (v) can be selected from a plurality of predetermined optimization values ​​(v).

8. Method according to claim 7, characterized in that the specified optimization values ​​(v) are based on the respective process parameter actual values ​​(p').

9. Method according to one of the preceding claims, characterized in that between the first time point (t1) and the second time point (ta) there is a first time interval (Δt1) comprising at least one magazine (d) and between the second time point (ta) and the third time point (t3) there is a second time interval (Δt2) comprising at least one magazine (d).

10. Method according to claim 9, characterized in that the first time interval (Δt1) and the second time interval (Δt2) have a different number of magazines (d), wherein the first time interval (Δt1) expediently has a single magazine (d).

11. Method according to one of the preceding claims, characterized in that a setpoint function (S) is stored in the control device (1) for each product property (w) to be controlled.

12. Method according to one of the preceding claims, characterized in that the setpoint function (S) for the at least one product property (w) to be controlled is formed from experimental data or from a setpoint process model.

13. Method according to claim 12, characterized in that the setpoint process model is based on a kinetic model of the at least one product property (w).

14. Method according to one of the preceding claims, characterized in that the at least one product property (w) to be controlled is recorded as a product property (p) measured at a first time (t1) and is transmitted to the control device (1) as the actual product property value (p ').

15. Method according to claim 14, characterized in that the recorded actual product property values ​​(p ') are smoothed by means of a smoothing method, preferably by means of the Whittaker-Henderson method.

16. Method according to claim 14 or 15, characterized in that the actual product property values ​​(p ') form a product property quantity set (p '').

17. Method according to one of claims 14 to 16, characterized in that the correction value (k) is calculated at the first time (t1) by subtracting from the process model product property value (w'c) of the at least one product property (w) calculated for the at least one product property (w) at the first time (t1) from the at least one product property actual value (w'm) recorded at the first time (t1).

18. Method according to any one of claims 14 to 17, characterized in that the acquisition of the product properties (w) is carried out as an inline measurement and / or atline measurement and / or online measurement, wherein the product property (w) is expediently measured with a product property sampling frequency (f w ) is recorded.

19. Method according to claim 4 and claim 18, characterized in that the process parameter sampling frequency (f P ) and the product property sampling frequency (f w) have the same value.

20. Method according to one of the preceding claims, characterized in that from a number of two product properties (w) to be controlled a product property set (w'') is formed, wherein a prioritization of the product properties (w) to be controlled is carried out with regard to a priority control of one of the product properties (w).

21. Method according to one of the preceding claims, characterized in that the at least one product property (w) is the particle size and / or the particle moisture and / or the particle composition.

22. Method according to one of the preceding claims, characterized in that the optimization model is based on the process model, in particular corresponds to the process model.

23. Method according to one of the preceding claims, characterized in that the process model for calculating the process model product property value (w'c) is based on a linear or nonlinear process model of the fluidization process to be controlled, wherein as For a nonlinear process model, it is expedient to use an artificial neural network.

24. Method according to claim 22, characterized in that the artificial neural network is designed as a multilayer perceptron or as a simple recurrent network or as a nonlinear autoregressive exogenous network.

25. Method according to one of the preceding claims, characterized in that one or more process parameters (p) from the group consisting of spray gas pressure and / or spray rate and / or spray quantity and / or particle temperature and / or drying gas temperature at the inlet of the fluidization apparatus and / or the relative humidity of the drying gas at the outlet and / or drying gas volume flow are used as process parameters (p).

Images