A method for predicting concrete slump based on current and sound signals during mixing

Abstract

The present invention relates to the field of concrete mixing technology, and in particular to a method for predicting the slump of concrete based on current and sound signals during mixing. The method comprises: S1, collecting current signals and sound signals; S2, data cleaning; S3, establishing a multimodal neural network; S4, predicting the slump of concrete. The present invention fuses the current signal collected during the concrete mixing process with the sound signal in the mixing cylinder, and performs online monitoring of the slump of concrete after feature extraction through a neural network. The slump monitoring method of the present invention can monitor the slump of concrete in mixing production in real time online under working conditions where the source of raw materials is complex and constantly fluctuating, thereby controlling the stability of concrete quality during production, improving production and construction efficiency, and avoiding waste of concrete materials due to unqualified slump.

Description

Technical Field

[0001] The invention relates to the technical field of concrete mixing, and in particular to a method for predicting concrete slump based on current and sound signals during mixing. Background Art

[0002] Concrete fluidity affects the quality and construction performance of concrete. Concrete slump is an important indicator to measure fluidity. Slump can be used to evaluate the working performance of concrete. Therefore, real-time monitoring of concrete slump during concrete production will effectively control the quality of concrete, improve on-site construction efficiency and avoid waste of resources caused by insufficient concrete working performance.

[0003] The traditional slump test is time-consuming and labor-intensive, has hysteresis, and due to sampling limitations, the concrete subjected to the slump test can only characterize the slump of a small part of the concrete. On-site operators usually observe the current curve during concrete mixing or the video of concrete mixing to judge the size of the concrete slump, but there are not many experienced operators and there is human subjectivity. At the same time, the source of raw materials for concrete mixing plants is complex, and the grading of aggregates and the moisture content of fine aggregates are difficult to control, which will make the slump of concrete even more difficult to control.

[0004] In order to solve these problems, concrete mixing stations urgently need to introduce a monitoring method for predicting the slump of concrete through current and sound during the mixing process, so as to monitor the slump of the mixed concrete and make timely formula adjustments. The present invention combines the current signal and sound signal collected by the current sensor and the sound sensor respectively, extracts and fuses the features of the two signals through an optimized neural network, and can realize real-time prediction of the slump of the mixed concrete. Summary of the invention

[0005] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or be understood by implementing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the description and other drawings of the description.

[0006] The purpose of the present invention is to overcome the above-mentioned shortcomings and provide a method for predicting the slump of concrete based on the current and sound signals during mixing. The current signal collected during the concrete mixing process and the sound signal in the mixing cylinder are integrated, and the slump of the concrete is monitored online after feature extraction through a neural network. The slump monitoring method of the present invention can be used to monitor the slump of concrete in mixing production in real time online under working conditions where the source of raw materials is complex and constantly fluctuating, thereby controlling the stability of the concrete quality during production, improving production and construction efficiency, and avoiding waste of concrete materials due to unqualified slump.

[0007] The present invention provides a method for predicting concrete slump based on current and sound signals during mixing, comprising:

[0008] S1. Collecting current signals and sound signals: a current collection system is set on the concrete mixer to collect the current signal of the main shaft of the mixer; a sound collection system is set inside the concrete mixer to collect the sound signal inside the mixer; the current signal and the sound signal are collected synchronously to form multi-modal information;

[0009] S2, data cleaning: perform data cleaning and normalization on the collected multimodal information to form two one-dimensional backbone networks;

[0010] S3. Establish a multimodal neural network: construct a multimodal neural network based on a one-dimensional convolutional neural network and a one-dimensional residual neural network, perform a feature superposition algorithm on the two obtained one-dimensional backbone networks, input them into the multimodal neural network, train the multimodal neural network, learn the characteristics of multimodal data, and form a mature multimodal neural network model;

[0011] S4. Predict concrete slump: Import the mature multimodal neural network model into the computer. When the mixer is running, read the current signal and sound signal in real time, import them into the mature multimodal neural network model, and the concrete slump can be predicted.

[0012] In some embodiments, in step S1, in order to make the output length and dimension of the current signal and the sound signal the same, a one-dimensional backbone network is constructed to perform feature extraction on the current signal and the sound signal, and the output length and dimension of the two one-dimensional backbone networks are the same.

[0013] In some embodiments, in step S3, a one-dimensional convolutional neural network is used to perform deep feature extraction on the superimposed signal data, and the one-dimensional convolutional neural network is composed as follows: the first layer is a convolution layer, the second layer is a global average pooling layer, the third layer is a convolution layer, the fourth layer is a global average pooling layer, and the fifth layer is an activation function layer.

[0014] In some embodiments, in step S3, the one-dimensional convolutional neural network constructs layers by stacking three residual blocks, each residual block contains two convolutional layers, each convolutional layer is connected to a batch normalization layer, ReLU is used as the activation function, the loss function is the mean absolute error mae, and the optimizer is the SGD algorithm.

[0015] In some embodiments, in step S3, the mature multimodal neural network model maps multiple results to the current signal and the sound signal after feature superposition, and adds a slump label to each result, thereby obtaining the corresponding slump of the input current signal and the sound signal.

[0016] In some embodiments, in step S4, it is necessary to identify the current signal and sound signal collected in real time. For the current signal, if the current signal does not converge over time but oscillates up and down, the current signal is judged to be an abnormal signal; for the sound signal, the time domain signal of the sound is Fourier transformed, and if the center frequency band deviates, the sound signal is judged to be an abnormal signal.

[0017] In some embodiments, in step S1, the starting signal of signal collection is determined by the feed door. When the feed door is opened, the mixer starts to feed and collects current signals and sound signals. The ending signal of signal collection is determined by the discharge door. When the discharge door is opened, the mixer stops working and stops collecting current signals and sound signals.

[0018] In some embodiments, in step S1, the current acquisition system includes a current acquisition module and a data acquisition card. The current acquisition module is electrically connected to the mixer motor through a current clamp, and the current signal is transmitted to the computer through the data acquisition card and saved. The sound acquisition system is provided with a microphone, which is arranged at the window of the mixer. The microphone transmits the sound signal to the computer and saves it.

[0019] In some embodiments, each batch of concrete produced generates a slump value after prediction. If the slump value meets the production requirements, the formula of the next batch of concrete will not be adjusted; if the slump value is too large, the amount of water is reduced when adding materials and the slump of the concrete is lowered; if the slump value is too small, the amount of water is increased when adding materials and the slump of the concrete is increased.

[0020] The invention discloses a device for predicting the slump of concrete based on the current and sound signals during mixing, comprising a mixer, a current collection system, a sound collection system and a computer terminal. The mixer is provided with the current collection system and the sound collection system, and the output ends of the current collection system and the sound collection system are connected to the computer terminal. The computer terminal is used to execute the method for predicting the slump of concrete based on the current and sound signals during mixing.

[0021] By adopting the above technical solution, the beneficial effects of the present invention are:

[0022] The present invention fuses the current signal collected during the concrete mixing process with the sound signal in the mixing cylinder, and performs online monitoring of the slump of the concrete after feature extraction through a neural network. The slump monitoring method of the present invention can perform real-time online monitoring of the slump of concrete in mixing production under working conditions where the source of raw materials is complex and constantly fluctuating, thereby controlling the stability of concrete quality in production, improving production and construction efficiency, and avoiding waste of concrete materials due to unqualified slump.

[0023] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure.

[0024] Undoubtedly, these and other objects of the present invention will become more apparent after the following detailed description of the preferred embodiment described with reference to various figures and drawings.

[0025] In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, one or several preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used together with the embodiments of the present invention to explain the present invention and do not constitute a limitation of the present invention.

[0027] In the drawings, the same reference numerals are used for the same components and the drawings are schematic and not necessarily drawn to scale.

[0028] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only one or several embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on such drawings without paying creative work.

[0029] Figure 1 It is a schematic diagram of the overall process of predicting slump in some embodiments of the present invention;

[0030] Figure 2 It is a schematic diagram of the overall structure of a slump prediction device in some embodiments of the present invention;

[0031] Figure 3 Schematic diagram of the structure of a one-dimensional network backbone and a multimodal neural network in some embodiments of the present invention.

[0032] Description of main reference numerals:

[0033] 1. Mixer; 2. Microphone; 3. Data acquisition card; 4. Computer terminal; 5. Control cabinet; 6. Current clamp; 7. Current signal input; 8. Sound signal input; 9. One-dimensional backbone network; 10. Feature superposition algorithm; 11. Multimodal neural network; 12. Slump output. DETAILED DESCRIPTION

[0034] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention, but are not used to limit the present invention.

[0035] In addition, in the description of the present invention, it should be understood that the terms "center", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "axial", "radial", "circumferential" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present invention.

[0036] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral body; it can be directly connected or indirectly connected through an intermediate medium, and it can be the internal connection of two elements or the interaction relationship between two elements. However, if it is indicated as a direct connection, it means that the two connected bodies are not connected through a transition structure, but are connected to form a whole through a connecting structure. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to the specific circumstances.

[0037] In the present invention, unless otherwise clearly specified and limited, the first feature "on" or "under" the second feature may be that the first and second features are in direct contact, or the first and second features are in indirect contact through an intermediate medium. In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in an appropriate manner in any one or more embodiments or examples.

[0038] Reference Figure 1-Figure 3 , Figure 1 It is a schematic diagram of the overall process of predicting slump in some embodiments of the present invention; Figure 2 It is a schematic diagram of the overall structure of a slump prediction device in some embodiments of the present invention; Figure 3 Schematic diagram of the structure of a one-dimensional network backbone and a multimodal neural network in some embodiments of the present invention.

[0039] According to some embodiments of the present invention, the present invention provides a method for predicting the slump of concrete based on the current and sound signals during mixing, such as Figure 1 As shown, including:

[0040] S1. Collecting current signals and sound signals: a current collection system is set on the concrete mixer 1 to collect the current signal of the main shaft of the mixer 1; a sound collection system is set inside the concrete mixer 1 to collect the sound signal inside the mixer 1; the current signal and the sound signal are collected synchronously to form multi-modal information;

[0041] The starting signal of signal collection is determined by the feeding door. When the feeding door is opened, the mixer 1 starts to feed and collects the current signal and the sound signal. The ending signal of signal collection is determined by the discharging door. When the discharging door is opened, the mixer 1 stops working and stops collecting the current signal and the sound signal.

[0042] like Figure 2 As shown, the current acquisition system includes a current acquisition module and a data acquisition card 3. The current acquisition module is electrically connected to the mixer 1 motor through a current clamp 6, and the current signal is transmitted to the computer terminal 4 through the data acquisition card 3 and saved; the sound acquisition system is provided with a microphone 2, and the microphone 2 is arranged at the window of the mixer 1. The microphone 2 transmits the sound signal to the computer terminal 4 and saves it; the mixer 1 motor is also connected to the control cabinet 5, and the control cabinet 5 is used to receive program data and control the motor running speed. Therefore, the current clamp 6 is used to clamp the power line between the mixer 1 motor and the control cabinet 5 to realize the current signal acquisition.

[0043] In order to make the output length and dimension of the current signal and the sound signal the same, a one-dimensional backbone network 9 is constructed. The two backbone networks differ in convolutional layers and step sizes. Feature extraction is performed on the current signal input 7 and the sound signal input 8 to make the output length and dimension of the two one-dimensional backbone networks 9 the same.

[0044] S2, data cleaning: performing data cleaning and normalization processing on the collected multimodal information to form two one-dimensional backbone networks 9;

[0045] S3, establishing a multimodal neural network 11: constructing a multimodal neural network 11 based on a one-dimensional convolutional neural network and a one-dimensional residual neural network, performing a feature superposition algorithm 10 on the two obtained one-dimensional backbone networks 9, inputting them into the multimodal neural network 11, training the multimodal neural network 11, learning the features of the multimodal data, and forming a mature multimodal neural network 11 model; in order to adapt to the different signal lengths of the two modalities, the two backbone networks differ in the convolution layer and step size,

[0046] A one-dimensional convolutional neural network is used to extract deep features from the superimposed signal data. The one-dimensional convolutional neural network is composed of the following: the first layer is a convolutional layer, the second layer is a global average pooling layer, the third layer is a convolutional layer, the fourth layer is a global average pooling layer, and the fifth layer is an activation function layer;

[0047] like Figure 3 As shown, the one-dimensional convolutional neural network constructs layers by stacking three residual blocks, each residual block contains two convolutional layers, each convolutional layer is connected to a batch normalization layer, ReLU is used as the activation function, the loss function is the mean absolute error mae, and the optimizer is the SGD algorithm; the three residual blocks are used to extract shallow, medium and deep feature information respectively to form a nonlinear mapping. In addition, the strength of the concrete needs to be tested. Concrete of different strengths corresponds to specific mixer currents and mixing times. Therefore, it is necessary to train respective multimodal neural networks according to the different concrete strengths11.

[0048] The mature multimodal neural network 11 model maps multiple results to the current signal and the sound signal after feature superposition, and adds a slump label to each result, so as to obtain the corresponding slump output 12 for the input current signal and the sound signal.

[0049] S4, predicting concrete slump: importing the mature multimodal neural network 11 model into the computer terminal 4, when the mixer 1 is running, reading the current signal and the sound signal in real time, importing them into the mature multimodal neural network 11 model, and predicting the concrete slump;

[0050] During the real-time monitoring process, it is necessary to first obtain the strength information of the concrete. The strength information of the concrete is known in advance. The corresponding trained model is called according to the strength, and then the current signal and sound signal collected in real time are identified. For the current signal, if the current signal does not converge over time but oscillates up and down, the current signal is judged to be an abnormal signal; for the sound signal, the time domain signal of the sound is Fourier transformed. If the center frequency band deviates, the sound signal is judged to be an abnormal signal; if the current signal and the sound signal are normal, the data is processed and then imported into the one-dimensional backbone network 9 and the multimodal neural network 11 to obtain the predicted slump.

[0051] Each batch of concrete produced will produce a slump value after prediction. If the slump value meets the production requirements, the formula of the next batch of concrete will not be adjusted; if the slump value is too large, the amount of water will be reduced when adding materials to lower the slump of concrete; if the slump is too small, the amount of water will be increased when adding materials to increase the slump of concrete. In addition, the corresponding parameters to be adjusted include not only water, but also the amount of raw materials such as sand, admixtures, fly ash, cement, etc., which can be flexibly adjusted according to production needs.

[0052] A device for predicting the slump of concrete based on current and sound signals during mixing, comprising a mixer 1, a current collection system, a sound collection system and a computer terminal 4. The mixer 1 is provided with the current collection system and the sound collection system, and the output ends of the current collection system and the sound collection system are connected to the computer terminal 4. The computer terminal 4 is used to execute a method for predicting the slump of concrete based on current and sound signals during mixing.

[0053] It should be understood that the embodiments disclosed in the present invention are not limited to the specific processing steps or materials disclosed herein, but should be extended to equivalent substitutions of such features understood by ordinary technicians in the relevant field. It should also be understood that the terms used herein are only used for the purpose of describing specific embodiments and are not meant to be limiting.

[0054] The "embodiment" mentioned in the specification means that the specific features or characteristics described in conjunction with the embodiment are included in at least one embodiment of the present invention. Therefore, the phrases or "embodiment" appearing in various places throughout the specification do not necessarily refer to the same embodiment.

[0055] In addition, the described features or characteristics may be combined in one or more embodiments in any other suitable manner. In the above description, some specific details, such as thickness, quantity, etc., are provided to provide a comprehensive understanding of the embodiments of the present invention. However, those skilled in the relevant art will understand that the present invention can be implemented without one or more of the above specific details or can also be implemented using other methods, components, materials, etc.

Claims

1. A method for predicting concrete slump based on current and sound signals during mixing, characterized in that: include S1. Collecting current signals and sound signals: a current collection system is set on the concrete mixer to collect the current signal of the main shaft of the mixer; a sound collection system is set inside the concrete mixer to collect the sound signal inside the mixer; The current signal and the sound signal are collected synchronously to form multi-modal information; S2, data cleaning: perform data cleaning and normalization on the collected multimodal information to form two one-dimensional backbone networks; S3. Establish a multimodal neural network: construct a multimodal neural network based on a one-dimensional convolutional neural network and a one-dimensional residual neural network, perform a feature superposition algorithm on the two obtained one-dimensional backbone networks, input them into the multimodal neural network, train the multimodal neural network, learn the characteristics of multimodal data, and form a mature multimodal neural network model; S4. Predicting concrete slump: Importing a mature multimodal neural network model into a computer, reading current signals and sound signals in real time during the operation of the mixer, and importing them into the mature multimodal neural network model to predict the concrete slump; In step S1, in order to make the output length and dimension of the current signal and the sound signal the same, a one-dimensional backbone network is constructed to extract features of the current signal and the sound signal, and the output length and dimension of the two one-dimensional backbone networks are the same; In step S3, a one-dimensional convolutional neural network is used to extract deep features from the superimposed signal data. The one-dimensional convolutional neural network is composed of the following: the first layer is a convolutional layer, the second layer is a global average pooling layer, the third layer is a convolutional layer, the fourth layer is a global average pooling layer, and the fifth layer is an activation function layer; This one-dimensional convolutional neural network constructs layers by stacking three residual blocks. Each residual block contains two convolutional layers. Each convolutional layer is connected to a batch normalization layer. ReLU is used as the activation function, the loss function is the mean absolute error mae, and the optimizer is the SGD algorithm. In step S4, the current signal and sound signal collected in real time need to be identified. For the current signal, if the current signal does not converge over time but oscillates up and down, the current signal is judged to be an abnormal signal; for the sound signal, the time domain signal of the sound is Fourier transformed. If the center frequency band deviates, the sound signal is judged to be an abnormal signal.

2. The method for predicting concrete slump based on current and sound signals during mixing according to claim 1, characterized in that: In step S3, the mature multimodal neural network model maps the superimposed signal data to multiple results and adds a slump label to each result, thereby achieving the corresponding slump of the input current signal and the sound signal.

3. The method for predicting concrete slump based on current and sound signals during mixing according to claim 1, characterized in that: In step S1, the starting signal of signal collection is determined by the feed door. When the feed door is opened, the mixer starts to feed and collects current signals and sound signals. The ending signal of signal collection is determined by the discharge door. When the discharge door is opened, the mixer stops working and stops collecting current signals and sound signals.

4. The method for predicting concrete slump based on current and sound signals during mixing according to claim 1, characterized in that: In step S1, the current acquisition system includes a current acquisition module and a data acquisition card. The current acquisition module is electrically connected to the mixer motor through a current clamp, and the current signal is transmitted to the computer through the data acquisition card and saved. The sound acquisition system is provided with a microphone, which is set at the window of the mixer. The microphone transmits the sound signal to the computer and saves it.

5. The method for predicting concrete slump based on current and sound signals during mixing according to claim 1, characterized in that: Each batch of concrete produced will have a slump value after prediction. If the slump value meets the production requirements, the formula of the next batch of concrete will not be adjusted; If the slump value is too large, reduce the amount of water when adding materials and adjust the slump of concrete downward; if the slump value is too small, increase the amount of water when adding materials and adjust the slump of concrete upward.

Images