Soil compaction system, soil compaction method and method for determining absolute soil compaction values

Abstract

The invention relates to a soil compaction system for compacting soil (2), comprising: a soil compaction machine (1); a data processing unit (8); and a determination device (10) for determining at least one input variable which is a criterion for the behaviour of at least one component of the soil compaction machine during a working phase; wherein the data processing unit (8) has an AI (artificial intelligence) device (9); the data processing unit (8) has been trained during a training phase; the data processing unit (8) is trained to derive at least one soil compaction value from the at least one input variable with the aid of the AI device (9); and the soil compaction value is an absolute soil compaction value.

Description

[0001] Soil compaction system, soil compaction methods and methods for determining absolute soil compaction values

[0002] The invention relates to a soil compaction system for compacting soil and a method for determining absolute soil compaction values ​​using a soil compaction machine. Furthermore, the invention relates to a soil compaction system with an algorithm or method in which input variables can be compared with a database generated by analytical, numerical, experimental approaches, etc., from which an absolute degree of soil compaction can be determined. The invention also relates to a soil compaction machine with a compaction unit, wherein the compaction unit is trained using a training method.

[0003] Soil compaction machines are well-known. They are used particularly for compacting soils on construction sites or in landscaping, ensuring that the soils have the necessary load-bearing capacity after the compaction process. Examples of such machines include vibratory plates, vibratory rollers, and rammers. Soil compaction is often carried out by workers who, based on their knowledge and experience, compact the soil and ultimately determine the compaction quality is sufficient based on their observations. In practice, the number of compaction passes is often specified by the construction plans to ensure adequate compaction in every case. Furthermore, it is possible to determine the degree of soil compaction using suitable measuring methods and conclude that the soil has been sufficiently compacted to achieve the required load-bearing capacity.

[0004] The extent or degree of soil compaction is also increasingly the subject of technical standards that must be considered and adhered to in construction projects.

[0005] Various measurement methods are known for verifying the degree of soil compaction achieved through soil compaction. Commonly used methods in practice, independent of the specific soil compaction machine, include determining the dry density (pd) and the degree of compaction according to Proctor (Dp). r , of the dynamic soil deformation modulus E V d by the dynamic plate load test, the soil deformation modulus Ev or the bedding modulus ks by the static plate load test, the penetration resistance (“probing resistance”) in the soil by means of dynamic probing, the soil density by radiometric methods or comparable methods.

[0006] Furthermore, there are machine-bound methods for determining the degree of compaction, such as the "Full-covering Dynamic Compaction Control" (FDVK) used with rollers or various compaction assistance systems for vibratory plates, which provide a relative dynamic characteristic value, offer a kind of measurement operation or aim to support certain verification methods.

[0007] Common methods for verifying the achieved degree of soil compaction of an area utilize these common measurement methods or machine-based procedures and integrate them into a verification concept.

[0008] The machine-independent methods commonly used in practice to determine the degree of soil compaction, i.e., those not coupled with a soil compaction machine, require special measuring and analysis equipment; in addition to the costs for the purchase and maintenance of the equipment and its regular calibration, personnel are also required who are trained in the operation of the measuring and analysis equipment and in the procedure.

[0009] To carry out the measurements, the compaction work must be interrupted. The length of the interruption or work stoppage in the respective area depends on the measurement method. For example, several minutes are needed for a single measurement point using the lightweight falling weight device or radiometric methods, whereas the static plate load test, including setting up or repositioning the measuring device and conducting the test, requires approximately one hour; this is necessary to determine the dry density pd and the Proctor compaction coefficient Dp. r From the time the sample is taken at the construction site until the measurement result is available, it can easily take 24 hours due to the necessary laboratory tests.

[0010] If the measurements to verify compaction are carried out by an independent service provider, advance planning and scheduling are required; additional waiting times may also occur. This consequently reduces the flexibility of the construction project.

[0011] Since intermediate checks are time-consuming, compaction usually continues until the specified compaction requirements are considered met; this often leads to over-compaction and thus to unnecessary additional work. If the measurement nevertheless reveals that the required degree of compaction has not been achieved, the compaction work must be resumed.

[0012] With conventional machine-independent measurement methods, the degree of soil compaction is determined only at selected points and within a narrowly defined area segment; information about the achieved degree of soil compaction across the entire work area is lacking. For example, if a work area is rejected as insufficiently compacted based on point measurements, it may be impossible – depending on the number and distribution of the measurement points – to determine how localized or widespread the weak spot is and what remedial measures need to be taken.

[0013] Determining the dry density pd requires taking a sample from the compacted area; accordingly, the sampling point must then be refilled and compacted.

[0014] A disadvantage of radiometric methods is that handling the radiometric samples is difficult.

[0015] Only persons who possess a permit and have undergone extensive training are allowed to operate the equipment. Furthermore, strict legal requirements govern the transport and storage of the equipment.

[0016] Performing the static plate load test requires a significant amount of space to set up the measuring device and position the heavy counter-support needed to apply the force, such as a truck, wheel loader, excavator, or roller. Therefore, this method is not applicable in confined spaces or less accessible areas.

[0017] Furthermore, the common measurement methods are not universal, but are only particularly suitable for certain soil types. For example, DIN 18125 BI. 2 recommends specific methods for determining the dry density pd, depending on the soil type.

[0018] As an alternative to the machine-free measurement methods described above, machine-based measurement methods for determining the degree of soil compaction are also known and are explained below. These are measurement methods that directly utilize the soil compaction machine used to compact the soil.

[0019] In "Full-Area Dynamic Compaction Control" (FDVK), calibration of the measuring system by compacting a calibration field is required to establish a correlation between the dynamic measurement, derived from the roller's vibration behavior, and the degree of soil compaction. The degree of soil compaction must then be determined – usually by Dp. r or expressed as Ev2 - also determined via suitable, in particular machine-independent, measurement methods, such as the methods listed above.

[0020] Since FDVK requires permanent contact between the compaction machine and the ground, the method is only applicable to rollers, but not to compaction machines with a jumping action, such as vibratory plates or rammers.

[0021] Compaction assistance systems, such as those used with vibratory plates, provide dynamic, machine-dependent parameters. These parameters represent relative values ​​that depend on the specific compaction machine and the specific soil; they do not correlate with the degree of soil compaction (e.g., Dp). ror Ev2), which is determined using standard (machine-independent) measurement methods. While the characteristic value is determined across the entire area, it only allows for a comparative, relative statement about soil compaction: areas that are more compacted than others can be identified; however, a quantitative statement about the absolute degree of soil compaction is not possible based on this characteristic value.

[0022] These types of compaction assistance systems can only give the operator an indication of whether a further increase in soil compaction can be achieved with the respective compaction machine; they only provide a decision aid regarding whether to continue or stop the compaction work.

[0023] Even with compaction assistance systems that enable measurement operation with permanent ground contact, the determined parameters currently do not show sufficient correlation to a degree of soil compaction measured using common methods (e.g., Dp). r or EV2). Furthermore, due to the transition from compaction work with intermittent operation to stationary measurement operation, the characteristic value is no longer determined continuously and across the entire area, but only locally.

[0024] The methods for verifying that the required soil compaction has been achieved reflect the state of the art in determining the degree of soil compaction and must deal with the associated shortcomings, such as the methods described in Section 14 of the ZTV E-StB:

[0025] In method M1, due to the high cost of carrying out the standard measurement procedures, the degree of soil compaction of the area to be tested is only determined at a few random samples. Information about the overall degree of soil compaction across the entire area is therefore lacking.

[0026] Method M 2 currently requires calibration of the measuring system on the roller by compacting a calibration field. Other compaction machine types, such as vibratory plates or rammers, cannot participate in this comprehensive verification method due to the lack of correlation between their dynamic characteristics and the degree of soil compaction. Method M 3 performs soil compaction under the same conditions (machine operating parameters, soil type, moisture content, number of passes, etc.) as those used in a test compaction to achieve the required degree of soil compaction. Therefore, the achieved soil compaction is not determined, but rather the work procedure is documented. If changes in certain influencing factors go unnoticed, such as an increase in moisture content due to rising groundwater resulting from the compaction work, then, as in point II, a change in soil compaction will also go undetected.

[0027] The invention is based on the objective of providing a soil compaction system and a method for determining the degree of soil compaction across the entire area and continuously using a soil compaction machine during a compaction process with a high correlation to common soil compaction measurement methods.

[0028] The invention is solved by a soil compaction system with the features of claim 1. In addition, a training method for training a data processing unit of a soil compaction machine, a working method, a method for determining absolute soil compaction values ​​with a soil compaction machine, and a soil compaction machine are specified.

[0029] A soil compaction system for compacting soil is described, comprising a soil compaction machine, a data processing unit, a determining device for determining at least one input variable that is a criterion for the behavior of at least one component of the soil compaction machine during a working phase, and a database generated by analytical, numerical and / or experimental approaches, containing relationships between the at least one input variable and an absolute degree of soil compaction, wherein the data processing unit is designed to compare the at least one input variable with the database during the working phase of the soil compaction machine and to determine an absolute soil compaction value.

[0030] The data processing unit may include an AI (Artificial Intelligence) device that has access to the database, wherein the data processing unit was trained during a training phase such that the database was populated, and wherein the data processing unit is trained to derive the absolute soil compaction value from the at least one input variable using the AI ​​device.

[0031] During compaction, the behavior of the compaction machine-soil system depends on numerous influencing factors; among these is the degree of soil compaction (soil compaction value), which affects, for example, the kinematic, dynamic, kinetic, or energetic behavior of the compaction machine, as well as the acoustics of the compaction machine and the compaction process. Certain patterns or relationships can thus be identified in the behavior of the compaction machine-soil system that are characteristic of a specific degree of soil compaction – determined using standard measurement methods (see above).

[0032] The invention solves the problem of determining the degree of soil compaction using a soil compaction machine that acquires data (input variables), optionally processes it appropriately, and then compares it with a database containing relationships between the data (input variables) and the degree of compaction. The machine then determines the degree of compaction based on this data comparison. The relationship between the data and the degree of compaction can be represented in the database, for example, by specific data patterns to which the corresponding degrees of compaction are assigned. The database can be created, for example, using machine learning, but also by other methods employing analytical, numerical, or experimental approaches.

[0033] During the creation of the database using machine learning, a data processing unit employs machine learning algorithms to identify, based on training data, the relationships between specific patterns in system behavior and the corresponding degree of soil compaction, and stores these relationships in a model. This enables the data processing unit to continuously recognize these patterns during the compaction process, in the interaction between the compaction machine and the soil, and thus continuously indicate the corresponding degree of soil compaction. This results in absolute rather than relative soil compaction values ​​with high correlation to common soil compaction measurement methods and high reproducibility.

[0034] This means, for example, that no calibration of the process is required on the construction site for the respective compaction task.

[0035] "Absolute" soil compaction values ​​are understood to be numerical values ​​provided by a common soil compaction measurement method, such as the deformation modulus EV2, the dry density p. d , the Proctor compression ratio Dp r etc. These absolute soil compaction values ​​therefore each represent an absolute degree of soil compaction.

[0036] "Relative" soil compaction values ​​are numerical values ​​that depend on the specific compaction machine, soil type and condition, etc., such as the characteristic values ​​determined by currently existing compaction assistance systems for vibratory plate compactors. However, these relative soil compaction values ​​are not transferable to other situations.

[0037] The soil compaction value should therefore be an absolute soil compaction value that can be correlated with soil compaction values ​​from known soil compaction measurement methods, which were explained in detail above.

[0038] The soil compaction system can include a data acquisition unit, a data evaluation unit (as part of the data processing unit; on the machine or externally), as well as a storage unit and / or an output unit (e.g., transmitting unit, visualization unit).

[0039] The data processing unit can be located in various places. For example, the data processing unit, or its components, can be mounted directly on or attached to the soil compaction machine. Alternatively, the soil compaction machine can be connected to the data processing unit in a suitable manner. For instance, the data processing unit, or its components, can be located in the cloud or on a smartphone, an external computer, or similar device. The soil compaction system allows for a process that, in one variant, enables the selection of a reference measurement method from several common soil compaction measurement methods, and the degree of compaction is then evaluated based on this method. Selecting multiple reference methods is also possible.

[0040] The method is applicable to compaction machines, such as vibratory plates, vibratory rammers, vibratory rollers, and attachment compactors.

[0041] For example, the method of machine learning, which is described below, can be used to create the database.

[0042] The process distinguishes between a "training phase" with, for example, machine learning of the data processing unit and a "working phase" in which the determination of the degree of compaction is carried out during the compaction work based on the previously learned model.

[0043] The training of the data processing unit during the training phase is carried out using learning data sets that represent aspects of the system behavior "compaction machine - soil" and to which the degree of soil compaction is already assigned - i.e., data sets that anticipate the result sought later during the compaction work.

[0044] The following activities are carried out during the training phase:

[0045] Data generation for training the data processing unit

[0046] - Processing of the generated data

[0047] - Application of machine learning algorithms to the training datasets to train the data processing unit and store them in a model

[0048] During the work phase, several different input variables can be determined, and consequently, several soil compaction values ​​can be derived. This means that the different input variables can relate to different parameters, which are measured using different measurement principles. The different input variables can be used to derive a uniform type of soil compaction value. They can also be used to derive different types of soil compaction values, according to the different measurement principles explained above.

[0049] It is also possible to determine many values ​​for a single input variable during the operation phase and derive a corresponding number of soil compaction values. In this case, a parameter type is recorded for a single input variable. Many values ​​are continuously determined for this parameter, and numerous soil compaction values ​​are derived from them. In this way, for example, the area traversed by the soil compaction machine can be monitored and evaluated with regard to its degree of soil compaction.

[0050] When generating data for training the data processing unit, the input variables ("features") used to determine the degree of soil compaction during the compaction process are first defined. Some possible input variables are listed below as examples; structuring these variables is advantageous:

[0051] The input variable can be a parameter that changes depending on the current degree of soil compaction during a given compaction operation by the soil compaction machine and describes the current behavior of an operating system, where the operating system comprises the soil compaction machine and the soil currently being compacted by it. This parameter is subsequently also assigned to "Category B".

[0052] Furthermore, it is possible that the parameter does not change, changes negligibly, or changes only with a small gradient during a given compaction operation. This parameter is subsequently assigned to "Category A".

[0053] Category A:

[0054] Possible input variables that describe the structure of the "compaction machine - soil" system and that do not change, change negligibly, or change only with a slight gradient during the respective compaction work include, for example:

[0055] Compaction machine type and / or soil type and / or sieve curve or asphalt type and / or water content of the soil to be compacted (soil moisture) and / or asphalt temperature (when applying the method to asphalt compaction) and / or layer thickness of the material to be compacted and / or inclination of the surface to be compacted (slope) and / or weather data and / or aging condition of machine components, such as, for example, the condition of rubber buffers, V-belts, undersize (mass loss due to wear) etc. in the case of vibratory plates.and / or geographic position data (preferably 3-dimensional) of the compaction machine, for example: o to record the compacted soil layer o to link the degree of soil compaction with the position for documentation and / or visualization purposes and / or operator guidance o to link the measurement of the degree of soil compaction with the input variables for the purpose of training the data processing unit (data labeling for machine learning).

[0056] Category B:

[0057] Possible input variables that change during the respective compaction process, influenced by the degree of soil compaction, and describe the instantaneous behavior of the "compaction machine - soil" system, include, for example, the movement and / or deformation behavior of at least one part of the machine, recorded, for example, via acceleration sensors (translational and / or rotational acceleration), structural strains, relative movements of machine components, changes in machine position, and / or electrical data such as: electrical current consumption (electric motor, etc.), and / or electrical voltage, and / or the (e.g., temporal) progression of one or more of the above-mentioned variables, and / or gradients (e.g.,temporal derivatives) of one or more of the above-mentioned quantities and / or acoustic data of the compaction machine and its interaction with the ground and / or mechanical and / or fluid-mechanical and / or thermal quantities, such as forces and / or moments on the compaction machine, the power output of the drive unit, the fuel consumption of an internal combustion engine, the exhaust gas temperature of an internal combustion engine.

[0058] The procedure described here requires at least one input parameter of category B; if this input parameter is not only influenced by the degree of soil compaction, but also depends on the design of the "compaction machine - soil" system, it is advantageous to also use further input parameters from category A.

[0059] Example:

[0060] The movement behavior of the compaction machine will also depend on the type of compaction machine (e.g., vibratory rammer, vibratory plate, roller, and their sub-variants with regard to power, structure, mass, etc.); accordingly, it is advantageous to include the type of compaction machine as an input variable in addition to an input variable that describes the movement behavior of the compaction machine.

[0061] To create a training dataset, in addition to at least one input variable, the corresponding soil compaction degree is required. This degree will later serve as the target variable ("label") during the training phase, and the data processing unit will be trained on its determination. The soil compaction degree can be determined, for example, using one of the measurement methods explained above and later, or a comparable method. A training dataset is then created by linking the at least one input variable with its corresponding soil compaction degree.

[0062] The more training datasets are used and the higher their data quality, the better the result of the training phase and thus the prediction probability in the working phase usually becomes.

[0063] To generate the necessary data, soil compaction tests can be conducted, for example, in which appropriately filled soil layers are compacted and the input parameters and the degree of soil compaction are varied within the desired range. During the compaction process, the input parameters are continuously recorded, and subsequently, the degree of compaction of the soil layer is determined at selected locations—as frequently as possible—using the previously defined measurement methods. The degree of soil compaction can also be varied by taking measurements after different numbers of compaction passes. The required number of passes depends on the desired range of values ​​for the degree of soil compaction.

[0064] If the geographical position of the machine is also recorded during the compaction runs, a direct assignment of input variables and the measured soil compaction degrees is possible for the creation of the learning data sets.

[0065] If soil compaction is determined using several different measurement methods, multiple soil compaction degrees can be assigned to the input variables. The data processing unit can thus be trained to determine the soil compaction degree according to different methods.

[0066] After data generation, processing the input variables can be advantageous to increase training success, such as data filtering, and / or sensor offset compensation, and / or input variable correction for correlations and influences that are sufficiently well measurable or analytically. combined method of analytical or numerical processing and / or evaluation and evaluation using machine-learned models), and / or an application of window functions, and / or a data transformation, for example the performance of a fast Fourier transform (FFT) to evaluate an input quantity in the frequency domain.

[0067] The data preparation measures to be applied, and in what order, must be considered when training the data processing unit, depending on the respective input variable and the machine learning algorithm used.

[0068] An output device may be provided for displaying the soil compaction value. This output device may include a visualization unit to convey or display information to an operator. Additionally or alternatively, the output device may include a transmission unit to send the determined soil compaction value information to an external unit, such as a server. It is also possible for the output device to include a storage unit in which the determined soil compaction values ​​are stored.

[0069] A training method for training a data processing unit in a training phase using training datasets is described. The training datasets represent aspects of the system behavior of an action system, where the action system comprises a soil compaction machine and soil currently being compacted by it, and where the system behavior is defined by the interaction of a soil compaction machine in a working phase and soil compacted by the soil compaction machine during the working phase.According to the training procedure, the following steps are to be carried out during the training phase: generating data for training the data processing unit, whereby the data includes input variables that represent a parameter for an aspect of the system behavior; preparing the generated data and creating training datasets; applying machine learning algorithms to the training datasets to train the data processing unit and storing them in a model in an AI facility.

[0070] The training procedure has already been explained in detail above, so a more detailed description is unnecessary here to avoid repetition. A training dataset can be created by linking at least one input variable with a corresponding soil compaction degree.

[0071] The input variable(s) can be processed using suitable means after they have been acquired.

[0072] Different algorithms can be used in the training process, whereby the different algorithms may differ with regard to the parameters of the input variables, a compilation of the training data sets and a measurement of the degree of soil compaction.

[0073] The data processing unit can be trained by applying machine learning algorithms to previously generated training datasets. The data processing unit learns to recognize data patterns or relationships between the at least one input variable and the degree of soil compaction and stores these in a model.

[0074] A multitude of machine learning methods now exist, and their algorithms and architectures are constantly evolving. Depending on the input variables, different learning methods with different configurations may be suitable, which is why a specific learning method will not be discussed in detail here; as an example, artificial neural networks can be mentioned, which consist of a large number of artificial neurons that can be arranged in virtually any number of layers.

[0075] Training success can be determined, for example, using control datasets where input variables were also recorded and the corresponding soil compaction degrees measured. However, the data processing unit is only provided with the input variables to determine training success. Based on this, the learned model predicts the soil compaction degree, which is then compared to the previously independently measured soil compaction degree of the control dataset.A working method is described for compacting soil and measuring soil compaction values ​​during a working phase, comprising the steps of: operating a soil compaction machine to compact the soil; determining at least one input variable that is a criterion for the behavior of at least one component of the soil compaction machine during the working phase; and deriving at least one soil compaction value from the at least one input variable using a KL device; wherein the soil compaction value is an absolute soil compaction value.

[0076] During the work phase – that is, during the actual compaction process – the degree of soil compaction is the unknown quantity being sought. To determine the degree of soil compaction, the input variables are recorded and processed during the compaction process, analogous to the training phase. Based on these input variables and the model learned during the training phase, the data processing unit calculates the degree of soil compaction. Because the recording and processing of the input data is continuous, the degree of soil compaction is also continuously determined during the compaction process.

[0077] If the data processing unit has been trained with regard to several different methods for measuring the degree of soil compaction, such as the Proctor compaction test, the static plate load test and the dynamic plate load test (see also the above description of the different measurement methods), the corresponding soil compaction values ​​(e.g. Dp) can also be obtained. r , Ev2 and Evd) are determined and output. Likewise, the machine operator could then select, during the work phase, which measurement methods the data is evaluated and output according to.

[0078] If the corresponding geographical machine position is also recorded, it can be linked to the respective degree of soil compaction, thus creating comprehensive information on soil compaction.

[0079] If the data processing unit outputs the determined soil compaction degree, this can, for example, be communicated to the machine operator and / or used for documentation and / or visualization of the compaction work. A method for determining absolute soil compaction values ​​using a soil compaction machine equipped with a compaction control device is described, along with a training procedure of the type described above for training the compaction control device of the soil compaction machine, and a work procedure of the type described above for compacting soil and measuring soil compaction values ​​during a work phase.According to the procedure, the following steps are to be carried out: operating the soil compaction machine to compact the soil; determining at least one input variable that is a criterion for the behavior of at least one component of the soil compaction machine during the working phase; and deriving at least one soil compaction value from the at least one input variable using the KL device; wherein the soil compaction value is an absolute soil compaction value.

[0080] A soil compaction machine is described, comprising a data processing unit and a computer unit, the computer unit being trained using a training method of the type described above. Specifically, the computer unit is trained using the training data sets and is thus able to determine the respective degree of soil compaction based on the recorded input variables.

[0081] A computer program is specified, comprising commands which, when executed by a computer, cause it to perform the steps of the aforementioned procedures, in particular training procedures, operating procedures, and determination procedures.

[0082] The method described here for determining the degree of soil compaction can also be applied analogously to determining the degree of asphalt compaction. During the training phase, corresponding training datasets are created containing asphalt compaction degrees determined using standard measurement methods. The input parameters of the method must be adjusted accordingly (e.g., asphalt type, asphalt temperature, etc., instead of soil type, water content, etc.).

[0083] The method described here for determining the degree of soil compaction can also be applied analogously to determining the degree of compaction of paving stones. During the training phase, corresponding training datasets are created using paving stone compaction degrees determined with standard measurement methods. The input parameters of the method must be adjusted accordingly.

[0084] The data processing unit can be located on the compaction machine or spatially separated from it. For example, the input variables required for the process could be transmitted to an external data processing unit and evaluated there (e.g., cloud computing).

[0085] Instead of a single input variable, multiple input variables ("sensor fusion") can be used, which can improve the quality of the results. These input variables can be of different types (e.g., acceleration sensors, electrical current measurements, etc.) and / or of the same type (e.g., multiple accelerometers applied at different positions on the machine).

[0086] The sensors for detecting one or more input variables can be machine-applied or non-machine-applied (external).

[0087] The input variable(s) can originate from sensors and / or other sources, such as operator input, delivery notes, quality assurance documents, external data sources, machine-specific data sources, etc.

[0088] The compaction machine type does not necessarily have to be specified as an input variable to the data processing unit during the compaction process (in the work phase), but can also be recognized independently by the data processing unit. For example, the data processing unit could be trained to recognize the compaction machine type using machine learning.

[0089] The soil type and / or grading curve or asphalt type does not necessarily need to be specified as input to the data processing unit during the compaction process (in the work phase), but can also be detected independently by the data processing unit. For example, the data processing unit could be trained to recognize the soil type and / or grading curve or asphalt type using machine learning. The geographical location of the construction site could also be used to make an assumption regarding the soil type. Experience shows that the soil type is often dependent on the geographical region of the construction site.

[0090] The asphalt temperature can also be determined independently by the compaction machine, for example via an infrared camera.

[0091] During compaction, input parameters can be continuously recorded on the compaction machine and stored on the machine and / or externally. The processing of these input parameters and / or the determination of soil compaction levels can also be performed with a time lag compared to the compaction process.

[0092] Input parameters can be continuously recorded during the activation of the compaction machine – i.e., even outside of the compaction work; an evaluation or output of the compaction degree then only takes place during the compaction work, which can also be detected independently, for example via machine operating data.

[0093] Input variables can also be recorded at intervals, intermittently, depending on time and / or position.

[0094] Data analysis can be performed on a rolling, sliding basis. The interval length (data set length) of the input variable acquisition and / or processing, and the frequency of determining the degree of soil compaction, can also vary.

[0095] Individual activities or steps in the process may also be omitted and / or carried out in a different order and / or in parallel.

[0096] To generate training datasets for machine learning from at least one input variable and the degree of soil compaction, the principles of statistical design of experiments could also be applied; for example, a full factorial or fractional factorial design to reduce the experimental effort. Instead of data from compaction test series, data from calculations or simulations of the "compaction machine - soil" system could also be used to generate training datasets for machine learning from at least one input variable and the degree of soil compaction, provided that the compaction machine, the soil, and the interaction between the compaction machine and the soil during the compaction process can be represented with sufficient accuracy as a computational or simulation model.

[0097] Data from actual construction projects could also be used.

[0098] During the training phase, the data processing unit can be trained to predict specific values ​​or ranges of soil compaction levels, or whether a soil compaction level threshold is exceeded or fallen below – also with corresponding selection options for the machine operator, etc.

[0099] During the training phase, the data processing unit can also be trained using learning data sets containing soil compaction profiles, in order to, for example, take into account the up- or down-oscillation of the compaction machine when using machine movement behavior as an input variable, and thus improve the prediction of the current soil compaction level.

[0100] During the training phase, the data processing unit can learn not only one model, but also several models using machine learning: For the individual values ​​of an input variable or for the combinations of individual values ​​of several input variables, separate models are trained and stored in the data processing unit via separate groups of training data sets.

[0101] During the processing phase, the data processing unit can select the appropriate model and apply it to determine the degree of soil compaction, depending on the current value of the input variable or the current combination of values ​​of several input variables. This further improves the predictive accuracy of the degree of soil compaction.

[0102] Example:

[0103] The input variables used for the procedure to determine the degree of soil compaction are: the type of compaction machine, the type of soil, the direction of travel (forward, reverse) and the machine accelerations.

[0104] During the training phase, a group of training datasets can be created for each combination of compaction machine type, soil type, and travel direction. The data processing unit is trained separately for each group of training datasets, resulting in a unique model for each combination.

[0105] During the operating phase, the data processing unit can use the specific compaction machine type, the specific soil type to be compacted, and the current direction of travel (e.g., forward or reverse) as input for pre-selecting the model. This pre-selected model is then used to evaluate the instantaneous machine accelerations during compaction operation and determine the degree of soil compaction.

[0106] When using multiple models, it is conceivable not to explicitly train all the characteristics or values ​​of one or more input variables and their combinations in separate models; a kind of interpolation between "neighboring" models is also conceivable.

[0107] For example, the input variable "soil type" has a very diverse range of characteristics (described by sieve curve, grain size, etc.); if a soil type not explicitly trained is present during the work phase, the models of the nearest trained soil types could be used for parallel evaluation; the soil compaction degree of the present soil is then deduced from the predicted soil compaction degrees of the nearest models.

[0108] During the training phase of machine learning, a wide variety of learning methods and algorithms can be used. For example, acoustic signals for determining the degree of soil compaction could also be evaluated using speech recognition software and its algorithms.

[0109] The data processing unit can also be trained on multiple output variables (degree of soil compaction determined using different measurement methods).

[0110] The data processing unit can also continue to learn and further develop the model(s) during the working phase if suitable data sets from the corresponding input variables and the soil compaction levels determined using common measurement methods are available.

[0111] During the work phase, the degree of soil compaction can also be used or processed internally - without an output unit - for example, to control or regulate an autonomously operating compaction machine.

[0112] A combination is specified consisting of an input unit (sensors, external data input, etc.) for capturing or providing at least one input variable for the process, a data processing unit, and a storage unit and / or an output unit.

[0113] The database can be created, for example, through machine learning, but also through other methods that pursue analytical, numerical, experimental approaches, etc.

[0114] The data processing unit is accordingly trained in a training phase – e.g., using machine learning algorithms or the aforementioned methods – on training datasets to recognize correlations between specific patterns in the system behavior of the compaction machine and the material to be compacted, and the corresponding degree of compaction; the recognized correlations are stored in one or more models. The training datasets consist of one or more input variables that describe the system behavior and the corresponding degree(s) of compaction, determined using standard measurement methods.

[0115] The data processing unit is enabled by the database to recognize these patterns in the input variables during the interaction between the compaction machine and the material to be compacted during the compaction work, following the training phase, and thus to indicate the respective degree of compaction.

[0116] A high correlation between the density level predicted by the data processing unit and the density level determined using common measurement methods is achieved through a large number of training data sets.

[0117] The described method enables the determination of absolute soil compaction degrees via a measuring system on the compaction machine, with a high correlation to conventional measurement methods and without the need for additional, specialized measuring and analysis equipment. The high costs for purchasing, maintaining, and regularly calibrating such equipment are eliminated. Specially trained personnel are not required.

[0118] Compared to conventional measurement methods, this offers a significant time advantage: The degree of soil compaction is determined during the compaction process itself; interrupting the compaction process is no longer necessary. The waiting times associated with laboratory test results are eliminated.

[0119] Unnecessary work resulting from exceeding compaction requirements is avoided: If the degree of soil compaction determined during the compaction process is communicated to the operator, they can monitor the compaction progress and stop work when the required degree of compaction is reached; this increases the efficiency and quality of the compaction work (time, costs). Furthermore, the operator, the environment, and adjacent existing buildings are protected (noise and exhaust emissions, vibration, etc.); resource consumption (energy, machine wear, etc.) is reduced. The degree of compaction is determined not just at a few points, but across the entire area; this allows for an assessment and documentation of the achieved compaction and thus the load-bearing capacity of the entire area.

[0120] Furthermore, the comprehensive determination of the degree of compaction enables the development of assistance systems, such as a worker guidance system that specifically directs the machine operator to areas that are still insufficiently compacted.

[0121] Continuous data collection and analysis allows for the creation of a compaction history of the area, which, in the case of insufficiently compacted areas, makes it possible to better identify the causes and to select targeted measures for remediation.

[0122] No additional space is required for measuring instruments or auxiliary materials (e.g., counter-supports for applying force during the static plate load test).

[0123] The process can be trained on a wide variety of materials to be compacted and is therefore basically universally applicable.

[0124] Calibration of the process to the specific material to be compacted is not required on site (at the construction site).

[0125] The method is also applicable to compaction machines that do not have permanent ground contact during compaction work.

[0126] The process allows the operator to evaluate the degree of compaction using several different measurement methods.

[0127] The procedure streamlines the compaction process with regard to verification and documentation.

[0128] Before the compaction machine is used on real construction sites, the data processing unit undergoes a training phase. Using machine learning algorithms and training datasets, the unit learns to recognize correlations between specific patterns in the system behavior of the compaction machine and the material being compacted, and the corresponding degree of compaction. The identified correlations are stored in one or more models. During the operational phase, the data processing unit applies these models to the currently recorded input variables and thus determines the current degree of compaction.

[0129] A high correlation to common measurement methods is achieved by training the data processing unit on a large number of training datasets for the individual characteristics or values ​​of the input variable(s).

[0130] Coverage of a wide range of use cases on the construction site is achieved by training the data processing unit using training datasets that contain a wide range of characteristics or values ​​of the input variable(s).

[0131] The required input parameters for the process are recorded on the compaction machine and / or originate from external sources, operator input, etc.; additional measuring instruments are not required.

[0132] The data required to determine the degree of soil compaction are continuously recorded and evaluated during the compaction work.

[0133] The use of a storage unit and / or output unit enables the use of the determined compression level for documentation purposes and / or for assistance systems.

[0134] Machine learning algorithms can reveal relationships that are difficult or impossible to grasp analytically.

[0135] These and other advantages and features are described below using examples.

[0136] The accompanying characters provide further explanation. They show:

[0137] Fig. 1 shows a schematic diagram of a soil compaction system according to the invention; Fig. 2 shows a flowchart with a training procedure; and

[0138] Fig. 3 shows a flowchart of a work process.

[0139] Fig. 1 shows a highly schematic representation of a soil compaction system with a soil compaction machine 1, which can be moved over the soil 2 to be compacted. In the example shown, the soil compaction machine 1 is a vibratory plate compactor. A vibratory roller, a rammer, or similar equipment could also be considered a soil compaction machine 1.

[0140] The soil compaction machine has a soil contact plate 3 which is set into vibration by an unbalanced exciter 4. The vibrations are transmitted through the soil contact plate 3 into the soil 2 to be compacted.

[0141] The ground contact plate 3 and the vibratory exciter 4 together form a lower mass 5. A superstructure 6 is arranged above the lower mass 5, which may, for example, include a drive (not shown) for the vibratory exciter 4. This is usually an internal combustion engine. However, electric versions are increasingly known as soil compaction machines 1 or vibratory plates. In this case, at least one electric motor can be provided on the vibratory exciter 4, which is powered by electrical energy from a battery 7 that is part of the superstructure 6.

[0142] The upper mass 6 is connected to the lower mass 5 by vibration decoupling devices 6a, e.g. rubber buffers, and is therefore movable within certain limits relative to the lower mass 5, so that the strong vibrations arising at the lower mass 5 are only introduced to the upper mass 6 to a small extent.

[0143] A data processing unit 8 is provided on the upper mass 6, which can perform control tasks for the soil compaction machine 1. For example, the data processing unit 8 can have a receiver for a remote control (not shown) so that the soil compaction machine 1 can be remotely controlled. The data processing unit 8 can also collect and store information on the operation of the soil compaction machine 1 and send it to an external receiver (not shown). Likewise, the data processing unit 8 can also receive information from an external sender.

[0144] For the invention, it is relevant that the data processing unit 8 has a database that has been trained to derive current soil compaction values ​​based on one or more input variables or parameters. The data processing unit 8 can include a computer unit 9 or a machine learning device. The database can, for example, be part of the computer unit 9 or the machine learning device. The parameters can relate, in particular, to the system behavior of the soil compaction system, especially the behavior of one or more components of the soil compaction machine 1.

[0145] The soil compaction values ​​are so-called absolute soil compaction values, meaning values ​​that can be correlated with known or standardized measurement methods. This eliminates the need to separately determine the degree of soil compaction using standardized measurement methods after a soil compaction process, as is the case with current technology. Instead, the soil compaction machine 1 can determine the respective degree of soil compaction during the compaction process itself or derive it from the system behavior of the soil compaction machine 1.

[0146] The soil compaction machine 1 can be equipped with sensors 10 in a suitable manner, each capable of detecting input variables or parameters. The control unit 9 was previously trained in a training procedure so that it can process the values ​​detected by the sensors 10 and derive the current degree of soil compaction from them. Two sensors 10 are shown as examples in Fig. 1: one sensor 10 on the upper mass 6 and another sensor 10 on the lower mass 5. The sensors 10 can be, for example, accelerometers.

[0147] Figure 2 shows a training procedure for training the AI ​​unit 9. The AI ​​unit 9 is configured to provide a machine learning module. Such AI units or modules are known per se, so a more detailed description is unnecessary here.

[0148] The creation of the models (e.g., using machine learning) or the database does not have to be performed on the soil compaction machine 1, but can preferably be carried out on external computers. The created model or database is then transferred to the data evaluation unit and used there to determine the degree of soil compaction.

[0149] The training of the Kl unit 9 takes place primarily at the manufacturer of the soil compaction machine 1. Upon delivery of the soil compaction machine 1, the training process can be fully completed, allowing the user to operate the machine without having to perform any further training.

[0150] In step S1, the input parameters and the compaction measurement method are first defined by the manufacturer of the soil compaction machine 1. In particular, they determine suitable measurement parameters that are necessary for the respective standardized compaction measurement methods and that allow conclusions to be drawn about the degree of soil compaction.

[0151] The input parameters and compaction measurement methods are experimentally verified in steps S2 and S3.

[0152] Step S4 defines how the measured data is to be processed.

[0153] In step S5, suitable training and control datasets can be created. These training datasets are then fed into a learning unit (e.g., learning unit 9), which can apply machine learning algorithms to the datasets and store them as a model within the learning unit (step S6). Step S6 represents the actual training process.

[0154] The creation of the models or the database will preferably take place on external computers, and the models or the database will then be transferred to the data evaluation unit. In step S7, the learning success can be checked with control data sets. In particular, the output data of the KfW unit 9 can be compared with the results of conventionally applied, proven compaction measurement methods.

[0155] If the prediction probability of the Kl unit 9 is sufficient (step S8), the training procedure can be terminated in step S9. If, however, the prediction probability is not yet sufficient, adjustments can be made at various points in step S10.

[0156] Fig. 3 shows an example of the steps of a work procedure in a work phase during data evaluation on the soil compaction machine 1.

[0157] In this process, S31 input variables are acquired in one step. This can be done, in particular, with the aid of sensors 10, as shown in the example in Fig. 1.

[0158] Data processing takes place in step S32.

[0159] Subsequently, based on the data obtained, the respective or current degree of compaction can be determined in one step S33 using the model stored in the Kl unit 9.

[0160] The degree of compaction determined by the Kl unit can be stored and / or output in step S34.

[0161] To complete the description, various known methods for determining the degree of soil compaction are explained below. These methods have already been discussed above, particularly in relation to the state of the art. Further details are given below.

[0162] Several measurement methods are known for determining the degree of soil compaction. For example, a commonly used method in practice is the determination of the dry density pd [g / cm³]. 3 ] a soil sample taken from the compacted area. For this purpose, the volume and dry mass of the extracted soil sample are determined. Soil sampling can be carried out, for example (see DIN 18125-2), using: o the core sampler method, o the balloon method, o the sand replacement method, o the liquid replacement method, o the gypsum replacement method, or a comparable method.

[0163] To assess the achieved soil compaction, the dry density pd is often compared to the Proctor density p. pr [g / cm 3 ] related and the resulting degree of soil compaction Dp r [%p r ] = Pd / p Prx 100% compared with a specified minimum soil compaction level. For example, minimum requirements for Dp r for road construction in the “Additional Technical Contractual Conditions and Guidelines for Earthworks in Road Construction” (ZTV E-StB).

[0164] Determining the Proctor density p, which depends on the soil type pr This is done via the Proctor test according to DIN 18127 or the "modified (improved) Proctor test".

[0165] Another method is the determination of the dynamic soil deformation modulus E. V d [MN / m 2] by the dynamic plate load test, performed with the light falling weight tester (see "Technical Testing Regulations for Soil and Rock in Road Construction" (TP BF-StB), Part B 8.3) or with the medium falling weight tester (see "Technical Testing Regulations for Aggregates in Road Construction" (TP Gestein-StB), Part 8.2.1), or a comparable method. In ZTV E-StB 09, the dynamic plate load test was significantly strengthened; Table 10 contains guideline values ​​for assigning the Evd to Dp. r contain.

[0166] Another method is the determination of the soil deformation modulus Ev [MN / m²]. 2 ] or the bedding modulus ks [MN / m²] 3] by the static plate load test according to DIN 18134 or a comparable method. In the method according to DIN 18134, two soil loading cycles are carried out, whereby the deformation modulus Evi is determined from the pressure-settlement curve (settlement-normal stress diagram) of the first loading and the deformation modulus EV2 is determined analogously from the second loading. For example, minimum requirements for the deformation modulus Ev2 [MN / m²] exist for road construction. 2 ] included in the “Additional Technical Contractual Conditions and Guidelines for Earthworks in Road Construction” (ZTV E-StB).

[0167] Another method is the determination of the penetration resistance (“probing resistance”) in the soil by means of dynamic probing according to DIN EN ISO 22476-2 or according to the “Technical Testing Regulations for Soil and Rock in Road Construction” (TP BF-StB) Part B 15.1 or a comparable method.

[0168] Another method is the determination of soil density by radiometric methods, as described, for example, in the FGSV standard no. 743 of the Research Association for Roads and Transportation (FGSV) or in the "Technical Testing Regulations for Soil and Rock in Road Construction" (TP BF-StB) Part 4.3 or a comparable method.

[0169] Furthermore, machine-based methods for determining the degree of soil compaction are known.

[0170] A machine-based, area-wide, dynamic method used with rollers to determine the degree of compaction is called "Area-Wide Dynamic Compaction Control" (FDVK). In this method, a measuring device installed in the machine measures the interaction between the roller and the soil during compaction; for this purpose, an accelerometer is permanently attached to the drum. A dynamic measurement is derived from the drum's vibration behavior, which correlates with the stiffness and compaction of the soil. The correlation between the dynamic measurement and the degree of soil compaction, expressed as the static deformation modulus Ev2 or the degree of compaction Dp, is then established. rPrior calibration of the roller on a calibration field is required, necessitating measurements according to the measurement procedures described above. This procedure is described in the "Technical Testing Regulations for Soil and Rock in Road Construction (TP BF-StB) Part E 2".

[0171] In order to adequately capture the interaction between the roller and the ground, this method requires permanent contact between the roller and the ground; therefore, it is not applicable to soil compaction machines that lift off the ground during compaction, such as vibratory plates or vibratory rammers.

[0172] Vibratory plates are equipped with compaction assistance systems that determine a dynamic characteristic value during compaction (in jump mode) from a measurement of the machine accelerations - to record the interaction between the vibratory plate and the ground - as well as from data on the machine operating behavior; from the relative change of the dynamic characteristic value during compaction, the operator can draw conclusions about the progress of soil compaction.

[0173] Furthermore, compaction assistance systems are available for vibratory plates, offering the option of selecting a measurement mode as a separate machine operating state. In contrast to compaction work with intermittent operation, during measurement mode the vibratory plate is in a state of vibration with permanent ground contact – similar to the measurement conditions used in the FDVK (Full-Through-Valve Compaction) of rollers. A measurement is generated from the continuous interaction between the vibratory plate and the ground, which is intended to allow an assessment of the achieved degree of soil compaction.

[0174] Another type of assistance system for compaction machines aims to support method M 3 (see ZTV E-StB section 14.1.4 and the brief description below) for testing soil compaction by recording the number of passes and machine operating parameters during compaction. The number of passes is usually determined and visualized based on the position of the compaction machine or a mobile device (e.g., a smartphone) located by the operator or at the compaction machine. This allows for a comparison with the required number of passes necessary to achieve a specific soil compaction level, given the required machine operating parameters.

[0175] Various methods are known that can be used to prove the degree of soil compaction achieved.

[0176] Typically, soil compaction work is subject to requirements regarding the quality of compaction, the fulfillment of which must be verified. Sections 3, 6, and 8 to 11 of the ZTV E-StB (German Technical Regulations for Electrical, Electronic & Information Technologies) specify such requirements for soil compaction (degree of compaction Dp). r , deformation modulus Ev2) for typical construction projects; furthermore, section 14 of the ZTV E-StB also describes three methods for testing soil compaction, which demonstrates the relevance of the measurement methods listed above for construction projects:

[0177] Method M 1 - Procedure according to statistical test plan:

[0178] In method M 1, the quality requirements specified by the ZTV E-StB (German Technical Regulations for Electrical, Electronic & Information Technologies) or the construction contract are checked on a sample basis, with the location of the test points in the compaction field determined by a random selection procedure. Before the test is carried out, it is determined which test characteristics will be examined at the test points; according to TP BF-StB Part E 1 or ZTV E-StB, the following values ​​are considered as test characteristics: the degree of compaction Dp. r the deformation modulus Ev2, the ratio of the deformation moduli EV2 / EVI, the air void fraction n a

[0179] The arithmetic mean and standard deviation are calculated from the test results of the sample, on the basis of which the result of the compaction work is accepted or rejected.

[0180] Method M 2 - Procedure for applying area-wide dynamic measurement methods: In Method M 2, proof that the agreed degree of soil compaction has been achieved is provided by the "Area-wide Dynamic Compaction Control" (FDVK, description see above). A correlation exists between the dynamic measurement value of the roller and the degree of soil compaction (static deformation modulus Ev2 or degree of compaction Dp). r ) under the present conditions of the soil compaction work, must be demonstrated by calibrating the roller on a suitable calibration field before the soil compaction work.

[0181] According to the ZTV E-StB (German Technical Regulations for Electrical, Electronic & Information Technologies), method M 2 is particularly suitable for construction projects with high daily output and largely uniform soil composition. The comprehensive determination of the dynamic measurement value allows for systematic identification of weaknesses in the compaction area – unlike spot checks or random sampling tests such as the aforementioned method M 1.

[0182] Due to the required permanent contact between the compaction machine and the ground, the M 2 method is currently only permitted for rollers.

[0183] Method M3 - Procedure for monitoring the work process:

[0184] Method M 3 is based on a trial compaction of each soil type to be used; it is usually carried out at the beginning of construction on the construction site. The procedure is described in TP BF-StB Part E 3.

[0185] The trial compaction is used to determine the operating parameters, the appropriate fill height, and the number of passes required to achieve the desired soil compaction for the selected compaction equipment. A work instruction then specifies the soil type with its permissible moisture content range, the type of compaction equipment selected, the relevant operating parameters, the maximum permissible fill height in the uncompacted state, and the number of compaction passes. Compliance with the work instruction must be documented during the compaction work.

[0186] The soil compaction quality achieved during the trial compaction can be determined by measuring the degree of compaction Dp. r or from parameters from indirect methods according to ZTV E-

[0187] StB section 14.2.5, such as the static plate load test (Ev2, Evi, EV2 / EVI, ks) and the dynamic plate load test (E Vd) the settlement test with the Benkelman beam, the test of the probe resistance (ram or pressure sounding) or a settlement measurement after the individual compaction transitions (in rock etc.).

Claims

Claims 1. Soil compaction system for compacting soil (2), comprising a soil compaction machine (1); a data processing unit (8); a determination device (10) for determining at least one input variable that is a criterion for the behavior of at least one component of the soil compaction machine during a working phase; and a database generated by analytical, numerical and / or experimental approaches, containing relationships between the at least one input variable and an absolute degree of soil compaction; wherein the data processing unit (8) is configured to compare the at least one input variable with the database during the working phase of the soil compaction machine (1) and to determine an absolute soil compaction value.

2. Soil compaction system according to claim 1, wherein the data processing unit (8) comprises an AI device (9) that has access to the database; the data processing unit (8) has been trained in a training phase such that the database has been populated; the data processing unit (8) is trained to derive the absolute soil compaction value from the at least one input variable using the AI ​​device (9).

3. Soil compaction system according to one of the preceding claims, wherein several different input variables can be determined during the working phase and accordingly a soil compaction value or several soil compaction values ​​can be derived.

4. Soil compaction system according to one of the preceding claims, wherein during the working phase many values ​​for an input variable can be determined and correspondingly many soil compaction values ​​can be derived.

5. Soil compaction system according to one of the preceding claims, wherein the input variable is a parameter that changes depending on the current degree of soil compaction during a respective compaction operation by the soil compaction machine (1) and describes the current behavior of an action system; and wherein the action system comprises the soil compaction machine (1) and the soil (2) currently being compacted by it.

6. Soil compaction system according to one of the preceding claims, wherein the parameter does not change or changes negligibly or only with a small gradient during a respective compaction operation.

7. Soil compaction system according to one of the preceding claims, wherein an output device is provided for outputting the soil compaction value.

8. Training method for training a data processing unit (8) in a training phase with training data sets representing aspects of the system behavior of an action system, wherein the action system comprises a soil compaction machine (1) and soil (2) currently being compacted by it; and wherein the system behavior is defined by the interaction of a soil compaction machine (2) in a working phase and soil (2) compacted by the soil compaction machine (1) during the working phase; comprising the steps to be performed during the training phase: Generating data for training the data processing unit, where the data includes input variables that represent a parameter for an aspect of the system behavior; Processing the generated data and creating training datasets; Applying machine learning algorithms to the training datasets for Training the data processing unit (8) and storing it in a model in a computer facility (9).

9. Training method according to claim 8, wherein a learning data set is formed by linking the at least one input variable with an associated soil compaction degree.

10. Training method according to claim 8 or 9, wherein the input variables are processed after their acquisition.

11. Training method according to one of claims 8 to 10, wherein different algorithms are used; and wherein the different algorithms may differ with respect to: parameters of the input variables, compilation of the training data sets, measurement of the degree of soil compaction.

12. Working method for compacting soil (2) and for measuring soil compaction values ​​during a work phase, comprising the steps: Operating a soil compaction machine (1) to compact the soil (2); Determining at least one input variable that is a criterion for the behavior of at least one component of the soil compaction machine (1) during the working phase; and Deriving at least one soil compaction value from the at least one input variable using a KL device (9); wherein the soil compaction value is an absolute soil compaction value.

13. Determination method for determining absolute soil compaction values ​​with a soil compaction machine (1) comprising a compaction device (9), comprising a training method according to any one of claims 8 to 11 for training the compaction device (9) of the soil compaction machine (1); and comprising a working method according to claim 12 for compacting a soil (2) and for measuring soil compaction values ​​during a working phase, comprising the steps: Operating the soil compaction machine (1) to compact the soil (2); Determining at least one input variable that is a criterion for the behavior of at least one component of the soil compaction machine (1) during the working phase; and Deriving at least one soil compaction value from the at least one input variable using the KL device (9); wherein the soil compaction value is an absolute soil compaction value.

14. Soil compaction machine (1) with a data processing unit (8) with a computer unit (9), wherein the computer unit (9) is trained with a training method according to one of claims 8 to 11.

15. Computer program comprising instructions which, when the program is executed by a computer, cause the computer to perform the steps of a method according to any one of claims 8 to 11 or according to claim 12 or according to claim 13.

Images