Device for controlling a modification process of hygroscopic material

Abstract

A device (10) for controlling a modification process of hygroscopic material (15), the device comprising a control unit (20) and a modification chamber (11) for the hygroscopic material, the device being configured to control the modification process of the hygroscopic material by - measuring at least one process variable of the modification process at least during the modification; and - measuring at least one process variable of the hygroscopic material at least during the modification; the device comprising a neural network having an output layer E for output data comprising at least one of the values ​​calculated by the neural network: color gamut, hardness, strength, modulus of elasticity (MOE), modulus of rupture (MOR), compression, density, dimensional stability, fire resistance, rot resistance, fungus resistance, and termite resistance;characterized in that the device (10) is further configured to control the modification process of the hygroscopic material by calculating at least one intermediate control parameter by a neural network using at least the measured process variables as input parameters of the neural network, wherein the at least one intermediate control parameter comprises a moisture gradient and / or a quantity of microcracks; and to control the modification process using genetic algorithms and genetic programming based on the at least one intermediate control parameter determined by the neural network.

Description

Technical area

[0001] The invention relates to a method and a device for controlling a modification process, such as a thermomechanical modification process, of hygroscopic material. background

[0002] Hygroscopic materials are materials with the ability to absorb and retain moisture from the surrounding air. When relative humidity changes, a difference in vapor partial pressure causes the material to absorb or desorb moisture to reach equilibrium. Due to their properties, hygroscopic materials find application in many industrial sectors. Typical hygroscopic materials used in industry include wood, wood-plastic composites, some plant-based materials, and concrete, for example.

[0003] Increased environmental awareness has led to a growing demand for more sustainable options. Among other things, materials that are not processed with toxic chemicals allow companies to build more responsibly and dispose of them conveniently when needed. Wood, as a natural, durable, and hygroscopic material, offers an attractive and cost-effective solution to the demands of environmental sustainability. However, up to 80% of the wood harvested in developing countries currently ends up as waste or is burned. In particular, less than 5% of tropical forests are sustainably managed.

[0004] Current wood processing methods, such as traditional drying and pressing processes for hardwood production, have proven economically unattractive to investors seeking to increase the proportion of harvested wood used for the production of sustainable wood products such as buildings and furniture. In particular, some low-value tropical wood species were deemed unsuitable for such applications because it was not possible to achieve the quality level required for such products, including mechanical properties, dimensional stability, and surface quality. Furthermore, the processing time for these and some higher-value wood species is often far too long, even when using the most modern drying and compaction processes.

[0005] Thermomechanical modification processes, such as Thermomechanical Timber Modification (TMTM), enable the utilization of a larger proportion of harvested wood material. With this thermomechanical modification process, the properties of the hygroscopic material, such as compressive strength and stiffness, density, hardness, and dimensional stability, can be modified to suit the specific application of the material. One known thermomechanical modification process is disclosed in publication WO 2022 / 175585 A1.

[0006] However, the effective and optimal control of modification processes, such as the thermomechanical modification process mentioned above, is problematic due to the multitude of factors influencing the processes. Summary

[0007] The aim of the present invention is to provide a novel method for controlling a modification process for hygroscopic materials, which significantly shortens the process time and also achieves an improved quality level for modified hygroscopic materials. In particular, the aim of the present invention is to provide a method for controlling a modification process for hygroscopic materials, which significantly improves quality parameters such as material strength, surface hardness, and dimensional stability of the modified hygroscopic material, while significantly shortening the process time.

[0008] The aim of the invention is achieved in that, in the method according to the invention for processing hygroscopic material, a modification process for modifying the hygroscopic material is controlled using genetic algorithms and genetic programming on the basis of at least one intermediate parameter calculated by a neural network, wherein at least one measured process variable of the modification process and at least one measured process variable of the hygroscopic material are used as input parameters for the neural network, and the neural network is trained in such a way that it takes into account case-by-case non-linear dependencies of the various input parameters as a function of the material and its initial state as well as the desired properties to be achieved for the final product.

[0009] The measured process variables of the modification process include, but are not limited to, air temperature, air velocity, relative humidity of the air, air pressure, compression force, and compression rate.

[0010] The measured process variables of the hygroscopic material include, among others: humidity, temperature, moisture gradient, occurrence of microcracks, size of the parts to be modified and compression (thickness).

[0011] Other suitable additional input parameters for the neural network can also be used.

[0012] Preferably, the modification process controlled by the method according to the invention is a thermomechanical modification process of the hygroscopic material. Furthermore, in the method according to the invention, the moisture gradient of the hygroscopic material is preferably measured by electrical impedance spectroscopy (EIS), and the occurrence of microcracks in the hygroscopic material is preferably monitored by acoustic emission (AE).

[0013] More specifically, the method according to the invention is characterized by what is stated in independent claim 1, and the device according to the invention is characterized by what is stated in independent claim 15. Dependent claims 2-14 show some advantageous embodiments of the method according to the invention.

[0014] The advantage of the method and device according to the invention is that the influencing factors in a hygroscopic material, such as the moisture gradient values ​​and the amount of microcracks, can be kept within an acceptable level during its modification, thus achieving an appropriate overall quality for such a modified hygroscopic material, even if the process is significantly shorter than before. In particular, the ability of the hygroscopic material to absorb moisture from the air and the strength of the material are improved. For this reason, hygroscopic materials, such as low-quality tropical woods, can be used in more sustainable applications, e.g.in furniture or building products, instead of being incinerated or landfilled as is common today.

[0015] In addition, genetic algorithms and genetic programming offer further advantages in this type of application by solving many different problems in controlling the change process that cannot be solved with current control methods. These include, among others: - Techniques for monitoring the drying phase are only useful if the mutual and overall effects of the observed process values ​​on the changes in the properties of the material to be dried are known, - Hygroscopic materials such as wood are generally inhomogeneous natural materials, with thousands of species existing, all with different physical properties. Therefore, each of them has its own specific characteristics that should be taken into account during processing to obtain a final product with the desired properties. - When it comes to changing certain physical properties (surface hardness, strength, moisture, torsional stiffness, ageing resistance without chemical, environmentally harmful substances), even approximately, the task represents a theoretical optimization situation for which genetic algorithms (GA) and genetic programming (GP) methods and suitable online measurement technology offer more effective tools than conventional methods that use a single predetermined control program based on a mathematical model. Short description of the drawings

[0016] The invention will be described in more detail below with reference to the accompanying drawings, in which Fig. shows schematically a cross section through a modification chamber of a modification device used in an embodiment of a device according to the invention, and Fig. shows an embodiment of an artificial neural network for use in the present invention. Fig. shows a flow chart for an embodiment of the method according to the invention. Detailed description of some advantageous embodiments of the invention

[0017] In the method according to the invention, a modification process for hygroscopic material, such as a thermomechanical modification process, is controlled. In such a thermomechanical process, hygroscopic material such as wood is treated in a modification device, which typically comprises a modification chamber into which hygroscopic material is introduced. In this modification device, the hygroscopic material is modified by applying a process comprising several process phases such as humidification, drying, heating, and compression. The atmospheric conditions in the modification chamber, such as air temperature, relative humidity of the air, and, for example, the compression force, are varied according to an applied control program.The method used depends, among other things, on the type and quality of the hygroscopic material prior to modification, the size of the pieces to be modified, and the desired appearance and mechanical properties the hygroscopic material is to achieve during modification. Once the desired properties of the hygroscopic material to be processed have been achieved, it is moved outside the modification chamber for further processing, such as packaging, storage, and / or transport, to another internal production department, or to the material's end user.

[0018] An example of such a modification device 10 is shown in Fig. This comprises a modification chamber 11 into which the hygroscopic material 15 can be introduced when it is thermomechanically modified. In this example, the hygroscopic material to be modified is in the form of a stack, i.e., it comprises a plurality of pieces of such material arranged side by side and / or one above the other, so that between each or some rows and / or columns of pieces of hygroscopic material, there may or may not be one or more spacers. These spacers 16 are preferably cellular boards, plates, or stickers with cavities or channels through which the air blown into the modification chamber can flow.

[0019] The hygroscopic material used in the Fig. The material to be modified by the modification device 10 shown can be, for example, wood, wood-plastic composites, or a plant-based material. These materials are preferably designed such that they behave like wood during the modification.

[0020] Located in the modification chamber 11 is a compression device 12 comprising a first compression element 13 and a second compression element 14, between which the batch of hygroscopic material 15 to be modified is placed and by means of which it can be compressed during modification. The first compression element 13 and the second compression element 14 are platform-like elements having flat compression surfaces between which the material to be modified can be placed during modification. In this embodiment, the compression device has only two compression elements and can therefore be used to compress the material to be modified in the thickness direction of the pieces of hygroscopic material.However, in this and in some other embodiments of the method and the device, the compression device may also comprise further compression elements in order to compress the pieces of hygroscopic material in other directions as well, ie in the width and / or length direction.

[0021] In the Modification Chamber 11 of the Fig. The modification device 10 shown also includes heating means 18, a fan 19, and humidification means (not shown in the figure). The heating means 18 may be, for example, an electric heater, an oil heater, or a suitable biofuel heating device. The fan 19 is preferably an electrically driven fan, and the humidification means may comprise liquid spray and / or steam devices or apparatus.

[0022] The Fig. The modification device 10 shown also comprises measuring means for measuring various properties of the hygroscopic material. In particular, when applying the method according to the invention, the moisture gradient of the hygroscopic material to be modified is preferably measured by means of electrical impedance spectroscopy (EIS), and the number of microcracks in the hygroscopic material is preferably measured by means of acoustic emission (AE). Other sensors and measurements can also be used to determine the moisture gradient and the number of microcracks. These measurements are carried out at least during the thermomechanical modification, but can also be carried out before and / or after the thermomechanical modification. Online measurements can be carried out by providing suitable sensors connected to the control unit 20 of the modification device 10, e.g.Wirelessly via a Wi-Fi network or other suitable wireless data communication technology. Alternatively, wired connections can also be used.

[0023] Some other process parameters and physical quantities, such as temperature, humidity of the air and / or of the hygroscopic material to be modified, as well as the weight of the hygroscopic material to be modified in the modification device 10, can also be measured using suitable measuring sensors. Both online measuring devices / sensors and testing devices / instruments can be available in conjunction with the modification device 10. Separate laboratory measurements can also be carried out to verify the determined property values. The laboratory measurements are preferably carried out for properties of the hygroscopic material such as moisture content, hardness, and strength of the material. The control unit 20, to which the online measuring devices are connected and / or into which the additional input data is fed, is arranged such that it controls the various devices of the modification device, such asthe compression device 12, the heating device 18, the fan 19 and the humidification device, according to the control program executed in the control unit 10. Thus, the control unit 20 comprises computing means for executing the control program and a memory in which the control programs and the processed data are stored. Therefore, the computing means has the ability to control these devices so that the method according to the invention can be carried out, i.e., it has programs for executing the neural network as well as the genetic algorithms and the genetic programming in order to carry out the data processing according to the method of the present invention. The control unit can also comprise the necessary electronic circuits and components to control the actuators, such asthe fans and actuators of the compression device, and the measuring devices of the modification device, so that the modification process carried out by the modification device can be fully automatically controlled by the control unit 20.

[0024] The above-mentioned control program in the control unit 20 thus controls the thermomechanical modification process of the hygroscopic material according to the method according to the invention. In this embodiment, the method comprises at least the following method steps: - Measurement of temperature and humidity in the modification chamber, - measuring the moisture gradient of the hygroscopic material, preferably by electrical impedance spectroscopy (EIS), at least during the modification; - monitoring the amount of microcracks occurring in the hygroscopic material, preferably by acoustic emission (AE), at least during the modification, - Calculation of at least one intermediate control parameter by a neural network using at least the measured variables as input parameters of the neural network, - Control of the thermomechanical modification process using genetic algorithms and genetic programming based on the at least one intermediate control parameter determined by the neural network.

[0025] Electrical impedance spectroscopy (EIS) is the measurement of the impedance (alternating current resistance) of an object at different frequencies. The result is a frequency spectrum that provides information about the structure and properties of the object. Impedance spectroscopy is used in various applications, one of the most important of which is the study of biological substances, for example, in medicine. In the present invention, EIS technology is used to monitor the structure, moisture distribution, and moisture gradient of a hygroscopic material in real time.

[0026] The measurement setup for electrical impedance spectroscopy may include suitable electrodes that are applied to the surface of the material to be measured and used to perform the impedance measurements during the modification. A separate measuring device with the electrodes may be located in the modification chamber, which is then wirelessly connected to the control unit 20 of the modification device.

[0027] Acoustic emission (AE) can be used to monitor microcracks that can occur during drying. When a hygroscopic material such as wood develops cracks during drying, the first phenomenon observed is microcracking. The material emits sound at ultrasonic frequencies during the formation of microcracks. By monitoring acoustic emission and adjusting the modification process accordingly, macrocracks that compromise the quality of the final product can be avoided.

[0028] The arrangement for measuring sound emissions can, for example, comprise piezoelectric sensors that measure the ultrasonic waves generated by the microcracks in the hygroscopic material to be monitored. The sensors can be wirelessly connected to the control unit 20 to provide corresponding measurement data.

[0029] Genetic algorithms are heuristic optimization methods that mimic the evolutionary mechanisms of nature. They are suitable for tasks where the solution space is very large (e.g., large combinatorial problems) and even the approximate optimum is sufficient for the solution. With the rapid growth of computer computing power, the possible applications of genetic algorithms have expanded greatly in the last decade.

[0030] Genetic algorithms (GA) and genetic programming (GP) as methods for regulating and controlling the thermomechanical modification process improve the quality of the final product. This is because traditional mathematical analysis does not, or cannot, provide an analytical solution. GA and GP are adaptive methods that adapt to changes in the modification process. The process is improved without external intervention through real-time optimization or machine learning. Adaptation occurs continuously. The internal relationships between the relevant variables are currently poorly understood (or there is a reasonable suspicion that the current understanding is incorrect). Finding the size and shape of the final solution to a problem is an important part of the problem. An approximate solution is acceptable (or is the only solution likely ever to be achieved).Such tasks generate a large amount of computer-readable data that must be reviewed, classified, and aggregated. Small performance improvements are routinely measured (or are easily measurable) and are significant to the overall outcome.

[0031] A further added value of genetic algorithms and genetic programming in the measurement and control technology of the thermomechanical modification process is that the focus during the drying of the hygroscopic material is not on controlling the humidity and temperature. Instead, the hygroscopic material to be dried is not damaged by the drying process, and the final moisture content of the dried hygroscopic material is sufficiently uniform, thus ensuring a low moisture gradient. This results in a processed hygroscopic material with an intact structure and the lowest possible moisture distribution, gradient, and desired moisture content.

[0032] To achieve this and other process goals, the EIS and AE monitoring methods presented above enable real-time detection of the time and conditions at which deterioration of the hygroscopic material begins. Knowing the time and circumstances allows efforts to be made to prevent such situations. This creates the conditions necessary to prevent the hygroscopic material from cracking or becoming damaged.

[0033] An artificial neural network (hereinafter “neural network”) as described in Fig. The neural network shown is a nonlinear statistical data modeling or decision-making tool. It can be used to model complex relationships between inputs and outputs or to find patterns in data. It is a network of simple processing elements (artificial neurons) that can exhibit complex global behavior determined by the connections between the processing elements and the element parameters. In the present invention, the neural network is used to determine the relationships between the properties of the hygroscopic material and the measured humidity gradient and microcracks, as well as other measured quantities of the hygroscopic material to be modified.

[0034] In the embodiment of Fig. The measurable input data for input layer A may include, but is not limited to: initial moisture content of the material; process temperature; relative humidity; steam consumption; moisture gradient; weight and / or density of the material; microcracking; wood species; compression pressure, rate and / or velocity; air velocity and / or direction; and / or changes in the input values.

[0035] In the neural network, the complex relationships between the input data and the output data of the output layer E are defined by the hidden layers BD, i.e., the deep learning phase, where the output data may include, among others: color scale; hardness; strength; elastic modulus (MOE); modulus of rupture (MOR); compression; density; dimensional stability; fire resistance; decay resistance; fungal resistance; termite resistance; calibration and compensation of temperature / humidity of process sensors.

[0036] Intermediate control parameters can be values ​​for the moisture gradient and microcracks calculated using the neural network based on the initial state of the hygroscopic material to be modified. They can also be other values ​​and / or combinations of values ​​calculated using the neural network and then used as inputs to the genetic algorithm used to control the modification device.

[0037] A flowchart of an embodiment of the method according to the invention is shown in Fig. shown.

[0038] In the embodiment of Fig. The modification process of the hygroscopic material is controlled based on available data, as shown in Box 101. At the beginning of the modification process, this data may be based on or include measurable variables related to the material being processed, such as moisture content, weight, density, wood species, and the process itself, such as process temperature, air velocity, compression pressure.

[0039] During the change process, data about the process is collected using appropriate sensors and measurements, as shown in Box 102.

[0040] The obtained process data are used as input data for the neural network, as shown in Box 103. Then, the output data of the neural network are used as input data for the genetic algorithm (see Box 104).

[0041] The genetic algorithm produces improved control data for the change process (see Box 105), which are then used to control the actual change process (see Box 101).

[0042] The Fig. The process described can be carried out as often as necessary until the desired quality and properties of the modified hygroscopic material are achieved with the modification process.

[0043] An embodiment of the method according to the present invention may further comprise determining target values ​​for the moisture gradient and the number of microcracks, as well as training the neural network and controlling the modification process with the aim of bringing the measured values ​​of the moisture gradient and the number of microcracks as close as possible to the target values ​​(i.e. values ​​calculated by means of the neural network).

[0044] An embodiment of the method of the present invention may further comprise determining the initial state of the hygroscopic material to be modified. Determining the initial state of the hygroscopic material refers to determining its properties, such as initial moisture content, initial moisture gradient, and initial amount of microcracks prior to modification. These can be determined by suitable tests and / or measurements. Measurements to determine the initial state of the hygroscopic material can be performed before modification as laboratory measurements or in the modification chamber before the start of modification using online measurements.Knowledge of the initial state of the hygroscopic material to be modified improves and accelerates the process and prevents situations where the control program does not achieve the best possible results due to deviations of the expected initial state from the actual initial state of the hygroscopic material.

[0045] An embodiment of the method of the present invention may further comprise the determination of occasional control values ​​and parameters. From case to case, some hygroscopic materials may have parameter values ​​that may deviate from the usual values. Some hygroscopic materials may also have a property or behavior due to which their modification requires some additional parameters to be used in controlling the process. In such a case, these parameters or values ​​may be determined before the start of the process or during the process, e.g., when a trigger value of a usual parameter or a combination of several values ​​of the usual parameters has been obtained.

[0046] An embodiment of the method of the present invention may further comprise drying the hygroscopic material using problem functions, variables, and parameters. Problem functions may be functions intended for the mathematical calculation of relationships between the measured or monitored parameters, such as the moisture gradient and the amount of microcracks, with respect to the atmospheric conditions used in the modification chamber or the pressing force of the compaction device.

[0047] An embodiment of the method of the present invention may further comprise evaluating the dried hygroscopic material. The evaluation of the dried hygroscopic material may include, for example, verifying the surface quality and dimensional accuracy of the pieces (such as straightness) of the hygroscopic material, measuring the strength and / or hardness, or determining the moisture content of the dried hygroscopic material.

[0048] An embodiment of the method of the present invention may further comprise the creation of a new control program based on the data obtained in the previous phases. When developing the new control program, the results of the neural network calculations can be used to obtain a control program that responds to changes in the measured parameters according to the desired properties sought for the specific hygroscopic material to be modified. In this way, the modification process can be controlled so that the hygroscopic material exhibits the properties most suitable for its intended use.

[0049] An embodiment of the method of the present invention may further comprise copying the best existing control program. Copying the best existing control program speeds up the calculation in the control unit 10 because, of course, if a more suitable starting point is chosen, the number of different phases and calculations required to achieve the optimal results is reduced by using the predetermined intermediate parameters.

[0050] An embodiment of the method of the present invention may further comprise the generation of a new control program through mutation. During mutation, a random part of a program is replaced by several other random parts of a program. In this embodiment, various combinations of replacements are tested through iteration to find the best possible solution for the particular case.

[0051] An embodiment of the method of the present invention may further comprise the creation of a new control program by cross-pollination. If the desired properties for the modified hygroscopic material lie somewhere between two or more existing control programs, a suitable control program can be achieved by combining the features of such existing control programs. The application of cross-pollination in such cases is sure to lead to a suitable control program more quickly, as fewer calculations are required than if the starting point were a single existing control program, which is farther from the final solution than a combination of two or more existing programs that already contain all or at least most of the required features.

[0052] An embodiment of the method of the present invention may further comprise selecting the best control program occurring in any population and using this control program in controlling the modification process. Such a method step is performed to find an optimal solution for each specific case in the application of genetic programming. As can be seen, there are various combinations of properties that can be achieved for a particular hygroscopic material. Furthermore, there are different types of hygroscopic materials and a variety of applications for these materials. Therefore, there are a large number of different combinations of control parameters that are best suited for all these different goals.Therefore, the results obtained with the different control programs must be evaluated and ranked according to their suitability in order to then choose which of the evaluated alternatives provides the best results for the specific product using the hygroscopic material chosen as raw material.

[0053] In one embodiment of the method of the present invention, the amount of water in the hygroscopic material is determined. The amount of water in the hygroscopic material can be determined using a microwave resonator or by measuring the initial weight of the hygroscopic material and the weight during the modification process, as described, for example, in the applicant's earlier publication WO 2022 / 175585 A1.

[0054] In one embodiment of the method according to the invention, the amount of water in the hygroscopic material is taken into account when calculating at least one intermediate control parameter by the neural network. The amount of water in the hygroscopic material influences the modification process via the moisture gradient, which must be determined at least during the modification process. Therefore, if the amount of water in the hygroscopic material is known, it is easier to predict the changes in the moisture gradient and the amount of microcracks occurring in the hygroscopic material. These are the main parameters describing the state of the hygroscopic material, which, in the present method, is to be optimized during its modification.

[0055] In some other embodiments of the process according to the invention, quality parameters other than moisture content, moisture gradient, and the number of microcracks may also be present. For example, if compression is used during modification, the density of the treated pieces can be determined by measuring the volume and weight of the pieces. Such additional control parameters governing the modification process allow the process to be further improved and, in particular, adapted for more specific uses and applications.

[0056] The method and device for controlling the modification process of the present invention are not limited to the embodiments described above, but can be varied within the scope of the claims. In further embodiments, one or more of the embodiments described above can be combined to obtain a suitable combination for achieving the desired properties of the hygroscopic material to be modified. QUOTES CONTAINED IN THE DESCRIPTION

[0000] This list of documents submitted by the applicant was generated automatically and is included solely for the convenience of the reader. This list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions. Cited patent literature

[0000] WO 2022 / 175585 A1 [0005, 0053]

Claims

[1] A device (10) for controlling a modification process of hygroscopic material (15), the device comprising a control unit (20) and a modification chamber (11) for the hygroscopic material, the device being configured to control the modification process of the hygroscopic material by - measuring at least one process variable of the modification process at least during the modification; and - measuring at least one process variable of the hygroscopic material at least during the modification; the device comprising a neural network having an output layer E for output data comprising at least one of the values ​​calculated by the neural network: color gamut, hardness, strength, modulus of elasticity (MOE), modulus of rupture (MOR), compression, density, dimensional stability, fire resistance, rot resistance, fungus resistance, and termite resistance; characterized byin that the device (10) is further configured to control the modification process of the hygroscopic material by calculating at least one intermediate control parameter by a neural network using at least the measured process variables as input parameters of the neural network, wherein the at least one intermediate control parameter comprises a moisture gradient and / or a quantity of microcracks; and to control the modification process using genetic algorithms and genetic programming based on the at least one intermediate control parameter determined by the neural network. [2] Device according to claim 1, characterized by that the at least one intermediate control parameter is calculated by the neural network using at least one measured process variable of the modification process and at least one measured process variable of the hygroscopic material as input parameters. [3] Device according to claim 1, characterized by that the at least one measured process variable of the hygroscopic material includes at least one of the following variables: the moisture gradient, the amount of microcracks. [4] Device according to claim 3, characterized by that the neural network has an input layer A for input parameters comprising at least one of the following parameters: the humidity gradient, the amount of microcracks, the relative humidity, the steam consumption, the weight of the hygroscopic material, the density of the hygroscopic material, the wood species, the compression pressure, the compression rate, the compression speed, the speed of the air and the air direction. [5] Device according to claim 4, characterized bythat the input layer A for the input parameters also comprises at least one of the following parameters: initial humidity of the hygroscopic material, process temperature, change in a value of an input parameter. [6] Device according to claim 4, characterized by that a quantity of water in the hygroscopic material (15) is taken into account when calculating the at least one intermediate control parameter by the neural network. [7] Device according to claim 6, characterized by that the amount of water in the hygroscopic material (15) is measured by a microwave resonator. [8] Device according to claim 4, characterized by that the device further comprises at least one of the following components: a compression device (12), heating means (18), fan (19).

Images