Self-adaptive compaction walking type road roller based on intelligent sensing and control method

Abstract

The invention discloses a self-adaptive compaction walking type road roller based on intelligent sensing and a control method, and relates to the technical field of walking type road rollers, the self-adaptive compaction walking type road roller comprises a rack, a walking rod and a road rolling roller; according to the method, multiple groups of real-time data and multiple groups of historical pressure data are acquired through the sensor group to calculate the compaction degree influence coefficient, meanwhile, the historical operation data of the road roller are acquired, and the compaction degree evaluation model is constructed on the basis of the convolutional neural network and in combination with the compaction degree influence coefficient to output the current compaction degree value and the uniformity index. A compaction state judgment result is obtained, real-time operation parameters of the road roller are adjusted in real time through a PID control algorithm, and the real-time operation parameters comprise the output power of a power system, the amplitude of a road rolling roller and the moving speed of the road roller.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of walk-behind road rollers, and particularly relates to a self-adaptive compaction walk-behind road roller based on intelligent sensing and a control method. BACKGROUND

[0002] The road roller is mainly a device for shaping and compacting asphalt concrete, ensuring the flatness of the asphalt concrete pavement, and reducing the wheel imprint of the pavement; it is mainly used for road maintenance and expansion, and with the development of the transportation industry, more and more roads need to be maintained and expanded. Small road rollers are small in size, sensitive in operation, and particularly strong in practicability in high road edge gaps and small side gaps, so the demand is also increasing.

[0003] During operation, the road roller needs to compact some narrow areas, at which time the large road roller is no longer applicable. In this case, a walk-behind road roller is selected, two compression rollers are installed below the machine body, the machine travels on the road surface relying on the compression rollers, and simultaneously performs the compaction operation. However, the traditional walk-behind road roller relies on manual operation experience to control the compaction parameters, has the risk of uneven compaction, overcompaction or undercompaction, and cannot realize real-time sensing of the working surface state.

[0004] In view of the above technical defects, the present application provides a solution. SUMMARY

[0005] The present application aims to: based on a convolutional neural network and combined with a compaction degree influence coefficient, a compaction degree evaluation model is constructed to output a current compaction degree value and a uniformity index, a compaction state judgment result is obtained, and a PID control algorithm is used to adjust the real-time operation parameters of the road roller in real time, including the output power of the power system, the amplitude of the rolling roller, and the moving speed of the road roller.

[0006] In order to achieve the above-mentioned purpose, the present application adopts the following technical scheme: a self-adaptive compaction walk-behind road roller based on intelligent sensing and a control method, comprising a rack, a handrail and a rolling roller, the handrail is fixedly arranged on the top surface of the rack, the inside of the rack is fixedly arranged with a power system, the rolling roller is sleeved on the outside surface of the power system, the outside surface of the rack is fixedly arranged with an extension seat, the outside surface of the extension seat is movably connected with an auxiliary support, and the outside surface of the auxiliary support is sleeved with a sensing roller shaft.

[0007] The sensing roller shaft comprises a roller shaft body and a wear-resistant layer, the roller shaft body is sleeved on the outside surface of the auxiliary support, and the wear-resistant layer is coated on the outside surface of the roller shaft body. A plurality of elastic studs and an electrically connected base are connected between the wear-resistant layer and the roller shaft body, and the inside of the electrically connected base is fixedly arranged with a sensor group.

[0008] Further, the auxiliary support comprises a lateral support and a limiting component, an outer side surface of the extension seat is fixedly provided with a vertical rail, the lateral support is movably connected to an inner wall of the vertical rail, and the limiting component is connected between the lateral support and the inner wall of the vertical rail, and a bottom end surface of the lateral support is fixedly provided with a driving motor, and an output end of the driving motor is connected to outer side surfaces of the roller shaft body.

[0009] Further, the limiting component comprises an electromagnet and a return spring, the electromagnet is fixedly provided on an outer side surface of the vertical rail, the return spring is connected between the lateral support and the inner wall of the vertical rail, and an inner part of the lateral support is filled with a magnet block.

[0010] Further, an outer side surface of the lateral support is fixedly provided with an extension support, a distal end of the extension support is connected with a cleaning scraper, and a side surface of the cleaning scraper is in contact with an outer side surface of the wear-resistant layer.

[0011] Further, the sensor group comprises a pressure sensor, a displacement sensor and a temperature sensor.

[0012] The application also provides a control method of a self-adaptive compaction hand-operated road roller based on intelligent sensing, comprising the following steps:

[0013] Step one, a plurality of sets of real-time data are obtained through the sensor group, a Kalman filtering algorithm is used to perform time-space alignment and noise suppression on the plurality of sets of real-time data, and a processed data set is obtained, the data set comprising temperature data, pressure data and displacement data;

[0014] Step two, a plurality of sets of historical pressure data are obtained to calculate a compaction degree influence coefficient, and historical operation data of the road roller are obtained, a compaction degree evaluation model is constructed based on a convolutional neural network and in combination with the compaction degree influence coefficient;

[0015] Step three, the data set is obtained, a real-time compaction degree influence coefficient is calculated according to real-time pressure data, the compaction degree influence coefficient, temperature data and displacement data are input into the compaction degree evaluation model to output a current compaction degree value and a uniformity index;

[0016] The displacement data are measured by a displacement sensor, and the displacement sensor is a linear displacement sensor, and since the displacement sensor is arranged between the wear-resistant layer and the roller shaft body, the measured data obtained are rotational displacement data Xi of the sensing roller shaft, and horizontal displacement data Xj of the sensing roller shaft are calculated according to the following formula: Further, the moving time of the sensing roller shaft is obtained, and the moving speed of the sensing roller shaft, i.e., the moving speed of the road roller, can be calculated in combination with the horizontal displacement data;

[0017] Step four, judging the current compaction state according to the preset compaction degree threshold and uniformity evaluation threshold, obtaining a compaction state judgment result, and adjusting the real-time operation parameters of the road roller in real time through a PID control algorithm, the real-time operation parameters including the output power of the power system, the amplitude of the rolling roller and the moving speed of the road roller.

[0018] Further, the specific process of calculating the compaction degree influence coefficient is as follows:

[0019] A data set is obtained, a plurality of groups of pressure data F and temperature data Ti are parsed from the data set, the force data F and the temperature data Ti are standardized and de-dimensioned, and the compaction degree influence coefficient Ki is calculated according to the following formula: Where d and e are preset weight coefficients, is a preset average pressure, i = 1, 2, 3, …, n, and the compaction degree influence coefficient is used to reflect the influence degree of the ground flatness to be processed and the environmental temperature on the compaction effect, the greater the compaction degree influence coefficient, the greater the influence degree of the ground flatness to be processed and the environmental temperature on the compaction effect, and vice versa, the smaller the compaction degree influence coefficient, the smaller the influence degree of the ground flatness to be processed and the environmental temperature on the compaction effect.

[0020] Further, the specific process of constructing the compaction degree evaluation model is as follows:

[0021] S201, the historical operation data of the road roller includes compaction process parameters under different material types, temperature data and initial compaction degree, and compaction degree influence coefficient, the historical operation data of the road roller is integrated as a training sample, and the training sample is split into a training set and a test set according to a proportion of 8:2;

[0022] S202, constructing a compaction degree evaluation model based on a convolutional neural network, downloading a weight file and loading it onto a corresponding network for initializing migration network parameters;

[0023] S203, modifying the last fully connected layer of the network, keeping the input unchanged, setting the output to the compaction degree value and the uniformity index, performing weight initialization on the last layer, using gradient descent algorithm for learning, and using fixed step attenuation to optimize the training parameters, retraining the entire network to obtain the compaction degree evaluation model;

[0024] S204, during the training process, a small batch of training samples are randomly and repeatedly extracted from the training set, all training samples are extracted as a training period, and the training is completed after a certain period of iteration to obtain the compaction degree evaluation model.

[0025] Further, the specific process of obtaining the compaction state judgment result is as follows:

[0026] S301, input the compaction degree influence coefficient and temperature data and displacement data into the compaction degree evaluation model to output the current compaction degree value Ai and the uniformity index Bi;

[0027] S302, obtain the preset compaction degree threshold Amin and the uniformity evaluation threshold Bmax, if the current compaction degree value Ai is greater than or equal to the compaction degree threshold Amin, and the uniformity index Bi is less than or equal to the uniformity evaluation threshold Bmax, the compaction state judgment result is excellent, and no parameter adjustment is needed;

[0028] If the current compaction degree value Ai is greater than or equal to the compaction degree threshold Amin, and the uniformity index Bi is greater than the uniformity evaluation threshold Bmax, the compaction state judgment result is good, and the parameter is adjusted in a small amplitude;

[0029] If the current compaction degree value Ai is less than the compaction degree threshold Amin, and the uniformity index Bi is less than or equal to the uniformity evaluation threshold Bmax, the compaction state judgment result is good, and the parameter is adjusted in a small amplitude;

[0030] If the current compaction degree value Ai is less than the compaction degree threshold Amin, and the uniformity index Bi is greater than the uniformity evaluation threshold Bmax, the compaction state judgment result is poor, and the parameter is adjusted in a large amplitude;

[0031] Further, the specific process of real-time adjustment of the real-time operation parameters of the road roller through the PID control algorithm is as follows:

[0032] The power system specifically includes a walking system and a vibration system for driving the road roller, and the walking system is specifically a driving wheel arranged below the machine body, and the driving wheel and the transmission system are used to realize the forward and backward movement of the device, and the operator can control the direction of forward or backward movement through the handle arranged on the handrail;

[0033] During the movement of the road roller, the ground is directly compacted by the rolling roller, and in order to improve the compaction effect, the hand-held road roller is also equipped with a vibration system, which makes the rolling roller vibrate through a mechanical device, so that the ground is more easily compacted during the compaction process;

[0034] S401, combine the BP neural network with the PID controller, wherein the input layer of the neural network is the multi-sensor data, the hidden layer approaches the optimal control law through self-learning, and the output layer directly adjusts the PID parameters;

[0035] S402, the PID parameter control process is as follows:

[0036] 1) Vibration frequency control: adjust the frequency through the vibration system (specifically a plunger pump displacement limiter), and the PID controller takes the error between the target frequency and the actual frequency as the input to dynamically adjust the hydraulic valve opening degree;

[0037] 2) Traveling speed control: using incremental PID algorithm (Δu(k) = P[e(k)-e(k-1)] + Ie(k) + D[e(k)-2e(k-1)+e(k-2)]), the output is the engine speed adjustment signal, where e is the walking efficiency of the walking system, and k is the efficiency influence coefficient;

[0038] 3) Amplitude and steering control:

[0039] Amplitude switching: according to the thickness of the layer, the corresponding amplitude is selected (i.e. high frequency and low amplitude for thin layer, low frequency and high amplitude for thick layer), and the hydraulic system pressure is adjusted by PID to realize rapid switching;

[0040] Steering control: based on the double GPS path planning information, the PID controller adjusts the steering angle, and the positive and negative transfer network algorithm is combined to ensure the tracking accuracy of the trajectory.

[0041] As described above, due to the adoption of the above technical solutions, the present application has the following advantages:

[0042] 1. The self-adaptive compaction hand-held road roller based on intelligent sensing and the control method, the frame is moved on the ground to be compacted by the hand-held rod, and then the ground is compacted by the power system driving the road roller, and the electromagnet is controlled to be powered off. Since the electromagnet loses magnetism, the magnetic attraction between the electromagnet and the magnet block disappears, the limiting effect of the contact side support is achieved, the side support falls vertically under the action of gravity, the wear-resistant layer is in advance contacted with the ground to be compacted, the roller shaft body is driven to rotate by the driving motor, the sensor group is continuously contacted with the ground to be compacted, the data is acquired by the sensor group to analyze the flatness and hardness of the ground to assist intelligent adjustment, and the wear-resistant layer is cleaned by the cleaning scraper to ensure the accuracy of the data.

[0043] 2. The control method of the self-adaptive compaction hand-held road roller based on intelligent sensing, a plurality of groups of real-time data are acquired by the sensor group, a plurality of groups of historical pressure data are acquired to calculate the compaction degree influence coefficient, historical operation data of the road roller are acquired, a compaction degree evaluation model is constructed based on convolutional neural network and combined with the compaction degree influence coefficient to output the current compaction degree value and uniformity index, the compaction state judgment result is acquired, and the real-time operation parameters of the road roller are adjusted in real time through the PID control algorithm. The real-time operation parameters include the output power of the power system, the amplitude of the road roller, and the moving speed of the road roller. BRIEF DESCRIPTION OF DRAWINGS

[0044] Figure 1 The overall external structure schematic diagram of the present application is shown;

[0045] Figure 2 Another angle of the overall external structure schematic diagram of the present application is shown;

[0046] Figure 3 A schematic diagram of the internal structure of the rolling roller of the present invention is shown;

[0047] Figure 4 A schematic diagram of the method flow of the present invention is shown;

[0048] Legend: 1. Frame; 2. Handrail; 3. Roller roller; 4. Extension seat; 5. Vertical track; 6. Lateral support; 7. Roller body; 8. Wear-resistant layer; 9. Elastic column head; 10. Power base; 11. Sensor group; 12. Drive motor; 13. Electromagnet; 14. Return spring; 15. Magnet block; 16. Extension support; 17. Cleaning scraper. Detailed Implementation

[0049] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] Example 1:

[0051] like Figures 1-3 As shown, the adaptive compaction walk-behind roller based on intelligent sensing includes a frame 1, a handrail 2, and a rolling roller 3. The handrail 2 is fixed to the top surface of the frame 1. A power system is fixed inside the frame 1. The rolling roller 3 is sleeved on the outer surface of the power system. An extension seat 4 is fixed to the outer surface of the frame 1. An auxiliary support is movably connected to the outer surface of the extension seat 4. A sensing roller shaft is sleeved on the outer surface of the auxiliary support.

[0052] The sensing roller includes a roller body 7 and a wear-resistant layer 8. The roller body 7 is sleeved on the outer surface of the auxiliary support. The wear-resistant layer 8 covers the outer surface of the roller body 7. Several elastic columns 9 and a power base 10 are connected between the wear-resistant layer 8 and the roller body 7. A sensor group 11 is fixed inside the power base 10.

[0053] The auxiliary support includes a lateral support 6 and a limiting component. A vertical rail 5 is fixed on the outer surface of the extension seat 4. The lateral support 6 is movably connected to the inner wall of the vertical rail 5. The limiting component is connected between the lateral support 6 and the inner wall of the vertical rail 5. A drive motor 12 is fixed on the bottom surface of the lateral support 6. The output end of the drive motor 12 is connected to the outer two sides of the roller body 7.

[0054] The limiting assembly comprises an electromagnet 13 and a return spring 14, the electromagnet 13 is fixed to the outer side surface of the vertical rail 5, and the return spring 14 is connected between the lateral support 6 and the inner wall of the vertical rail 5, and the inside of the lateral support 6 is filled with a magnet block 15.

[0055] The outer side surface of the lateral support 6 is fixed to an extension support 16, and the end of the extension support 16 is connected with a cleaning scraper 17, and one side surface of the cleaning scraper 17 is in contact with the outer side surface of the wear-resistant layer 8.

[0056] The sensor group 11 comprises a pressure sensor, a displacement sensor and a temperature sensor.

[0057] The working principle is as follows: the frame 1 is pushed on the ground to be compacted through the hand-held rod 2, and then the ground is compacted through the power system driving the road roller 3, and the electromagnet 13 is controlled to be powered off, the magnetic attraction between the electromagnet 13 and the magnet block 15 disappears due to the loss of magnetism of the electromagnet 13, the limiting effect of the lateral support 6 is lost, the lateral support 6 falls vertically under the action of gravity, and the wear-resistant layer 8 is in contact with the ground to be compacted in advance, the roller shaft body 7 is driven to rotate through the driving motor 12, the sensor group 11 is in contact with the ground to be compacted, the data of the sensor group 11 are acquired to analyze the flatness and hardness of the ground to assist intelligent adjustment, and the wear-resistant layer 8 is cleaned through the cleaning scraper 17, so that the accuracy of the data is ensured.

[0058] Embodiment 2:

[0059] As shown in Figure 4 ,

[0060] The application also provides a control method of a self-adaptive compacting hand-held road roller based on intelligent sensing, comprising the following steps:

[0061] Step one, a plurality of sets of real-time data are acquired through the sensor group 11, Kalman filtering algorithm is adopted to perform time-space alignment and noise suppression on the plurality of sets of real-time data, and a processed data set is obtained, the data set comprises temperature data, pressure data and displacement data;

[0062] Step two, a plurality of sets of historical pressure data are acquired to calculate a compaction degree influence coefficient, and historical operation data of the road roller are acquired, a compaction degree evaluation model is constructed based on a convolutional neural network and in combination with the compaction degree influence coefficient;

[0063] The specific process of calculating the compaction degree influence coefficient is as follows:

[0064] A data set is acquired, a plurality of sets of pressure data F and temperature data Ti are parsed from the data set, the pressure data F and the temperature data Ti are subjected to standardization and de-dimensioning processing, and then the compaction degree influence coefficient Ki is calculated according to the following formula: wherein d and e are preset weight coefficients, is a preset pressure average value, i = 1, 2, 3, …, n, and the compaction degree influence coefficient is used to reflect the influence degree of the ground flatness to be processed and the environmental temperature on the compaction effect, and the greater the compaction degree influence coefficient, the greater the influence degree of the ground flatness to be processed and the environmental temperature on the compaction effect, and vice versa, the smaller the compaction degree influence coefficient, the smaller the influence degree of the ground flatness to be processed and the environmental temperature on the compaction effect.

[0065] The specific process of constructing the compaction degree evaluation model is as follows:

[0066] S201, the historical operation data of the road roller includes compaction process parameters under different material types, temperature data and initial compaction degree, and a compaction degree influence coefficient, the historical operation data of the road roller is integrated as a training sample, and the training sample is split into a training set and a test set according to a proportion of 8:2;

[0067] S202, a compaction degree evaluation model based on a convolutional neural network is constructed, a weight file is downloaded and loaded onto a corresponding network for initializing migration network parameters;

[0068] S203, the last fully connected layer of the network is modified, the input is kept unchanged, the output is set to a compaction degree value and a uniformity index, the last layer is weight initialized, a gradient descent algorithm is used for learning, and fixed step attenuation is adopted to optimize training parameters, the entire network is retrained, and a compaction degree evaluation model is obtained;

[0069] S204, small batches of training samples are randomly and repeatedly extracted from the training set during the training process, all training samples are extracted as a training period, the training is completed after a certain period of iteration, and a compaction degree evaluation model is obtained.

[0070] Step three, obtain a data set, calculate a real-time compaction degree influence coefficient according to real-time pressure data, input the compaction degree influence coefficient, temperature data and displacement data into the compaction degree evaluation model to output a current compaction degree value and a uniformity index;

[0071] The displacement data is measured by a displacement sensor, and the displacement sensor is a linear displacement sensor. Since the displacement sensor is arranged between the wear-resistant layer and the roller shaft body, the measured data obtained is the rotation displacement data X1 of the sensing roller shaft. The horizontal displacement data Xj of the sensing roller shaft is calculated according to the following formula: Further, the moving time of the sensing roller shaft is obtained, and the moving speed of the sensing roller shaft, i.e., the moving speed of the road roller, can be calculated in combination with the horizontal displacement data.

[0072] Step four, judging the current compaction state according to the preset compaction degree threshold and uniformity evaluation threshold, obtaining the compaction state judgment result, and adjusting the real-time operation parameters of the road roller in real time through the PID control algorithm, the real-time operation parameters including the output power of the power system, the amplitude of the rolling roller 3 and the moving speed of the road roller.

[0073] The specific process of obtaining the compaction state judgment result is as follows:

[0074] S301, input the compaction degree influence coefficient, temperature data and displacement data into the compaction degree evaluation model to output the current compaction degree value Ai and the uniformity index Bi;

[0075] S302, obtain the preset compaction degree threshold Amin and the uniformity evaluation threshold Bmax, if the current compaction degree value Ai is greater than or equal to the compaction degree threshold Amin, and the uniformity index Bi is less than or equal to the uniformity evaluation threshold Bmax, the compaction state judgment result is excellent, and no parameter adjustment is needed;

[0076] If the current compaction degree value Ai is greater than or equal to the compaction degree threshold Amin, and the uniformity index Bi is greater than the uniformity evaluation threshold Bmax, the compaction state judgment result is good, and the parameter is adjusted in small amplitude;

[0077] If the current compaction degree value Ai is less than the compaction degree threshold Amin, and the uniformity index Bi is less than or equal to the uniformity evaluation threshold Bmax, the compaction state judgment result is good, and the parameter is adjusted in small amplitude;

[0078] If the current compaction degree value Ai is less than the compaction degree threshold Amin, and the uniformity index Bi is greater than the uniformity evaluation threshold Bmax, the compaction state judgment result is poor, and the parameter is adjusted in large amplitude.

[0079] The specific process of adjusting the real-time operation parameters of the road roller in real time through the PID control algorithm is as follows:

[0080] The power system specifically includes a driving system for the road roller and a vibration system, the driving system specifically is a driving wheel arranged below the frame 1, the driving wheel and the transmission system are used to realize the forward and backward movement of the device, and the operator can control the direction of forward or backward movement through the handle arranged on the handrail 2;

[0081] During the movement of the road roller, the ground is directly compacted by the rolling roller 3, and in order to improve the compaction effect, the hand-held road roller is also equipped with a vibration system, the vibration system makes the rolling roller 3 vibrate through a mechanical device, so that the ground is more easily compacted during the compaction process;

[0082] S401. Combine a BP neural network with a PID controller, where the input layer of the neural network is multi-sensor data, the hidden layer approximates the optimal control law through self-learning, and the output layer directly adjusts the PID parameters.

[0083] The S402 PID parameter control process is as follows:

[0084] 1) Vibration frequency control: The frequency is adjusted by the vibration system (specifically the piston pump displacement limiter). The PID controller uses the error between the target frequency and the actual frequency as input to dynamically adjust the hydraulic valve opening.

[0085] 2) Driving speed control: An incremental PID algorithm (Δu(k)=P[e(k)-e(k-1)]+Ie(k)+D[e(k)-2e(k-1)+e(k-2)]) is adopted, and the output is the engine speed adjustment signal, where e is the driving efficiency of the driving system and k is the efficiency influence coefficient;

[0086] 3) Amplitude and steering control:

[0087] Amplitude switching: Select the corresponding amplitude according to the layer thickness (i.e., high frequency and low amplitude for thin layers, and low frequency and high amplitude for thick layers), and achieve rapid switching by adjusting the hydraulic system pressure through PID control;

[0088] Steering control: Based on dual GPS path planning information, the PID controller adjusts the steering angle, and the forward and reverse transmission network algorithm ensures trajectory tracking accuracy.

[0089] This invention acquires multiple sets of real-time data through sensor group 11, acquires several sets of historical pressure data to calculate the compaction degree influence coefficient, and acquires the historical operating data of the road roller. Based on a convolutional neural network and combined with the compaction degree influence coefficient, a compaction degree evaluation model is constructed to output the current compaction degree value and uniformity index, obtain the compaction state judgment result, and adjust the real-time operating parameters of the road roller in real time through a PID control algorithm. The real-time operating parameters include the output power of the power system, the amplitude of the rolling roller 3, and the moving speed of the road roller.

[0090] The threshold is set to facilitate comparison. The size of the threshold depends on the amount of sample data and the number of bases set by those skilled in the art for each set of sample data; as long as it does not affect the ratio between the parameter and the quantized value, it is acceptable.

[0091] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0092] In the two embodiments provided in the present application, it should be understood that the disclosed apparatus and method can be implemented in other manners; for example, the apparatus embodiments described above are merely schematic; for example, the division of the modules is only a logical function division; there can be another division manner during actual implementation; for another example, the coupling or direct coupling or communication connection between the modules can be indirect coupling or communication connection through some interfaces, and mechanical or electrical form, or other form.

[0093] The above describes only the preferred embodiments of the present application, but the protection scope of the present application is not limited to this, any person skilled in the art can make equivalent replacements or changes to the technical solution and the inventive concept of the present application within the technical range disclosed by the present application, which should be covered in the protection scope of the present application.

Claims

1. A self-adapting compactor based on intelligent sensing, comprising a frame (1), a handrail (2) and a roller (3), characterized in that, The hand lever (2) is fixed to the top surface of the frame (1), the inside of the frame (1) is fixed with a power system, the rolling roller (3) is sleeved on the outside surface of the power system, the outside surface of the frame (1) is fixed with an extension seat (4), the outside surface of the extension seat (4) is movably connected with an auxiliary support, the outside surface of the auxiliary support is sleeved with an induction roller shaft; The induction roller shaft includes a roller shaft body (7) and a wear-resistant layer (8), the roller shaft body (7) is sleeved on the outside surface of the auxiliary support, the wear-resistant layer (8) is coated on the outside surface of the roller shaft body (7), a plurality of elastic studs (9) and an electrically connected base (10) are connected between the wear-resistant layer (8) and the roller shaft body (7), and a sensor group (11) is fixed in the electrically connected base (10).

2. The smart sensor based self-adaptive compactor ride-on roller according to claim 1, wherein, The auxiliary support includes a lateral support (6) and a limiting assembly, the outside surface of the extension seat (4) is fixed with a vertical rail (5), the lateral support (6) is movably connected to the inner wall of the vertical rail (5), the limiting assembly is connected between the lateral support (6) and the inner wall of the vertical rail (5), and the bottom end surface of the lateral support (6) is fixed with a driving motor (12), and the output end of the driving motor (12) is connected to the outer two side surfaces of the roller shaft body (7).

3. The smart sensor based self-adaptive compactor ride-on roller according to claim 2, wherein, The limiting assembly includes an electromagnet (13) and a return spring (14), the electromagnet (13) is fixed to the outside surface of the vertical rail (5), the return spring (14) is connected between the lateral support (6) and the inner wall of the vertical rail (5), and the inside of the lateral support (6) is filled with a magnet block (15).

4. The smart sensor based self-adaptive compactor ride-on roller according to claim 3, wherein, The outside surface of the lateral support (6) is fixed to an extension support (16), the end of the extension support (16) is connected with a cleaning scraper (17), and one side surface of the cleaning scraper (17) is in contact with the outside surface of the wear-resistant layer (8).

5. The smart sensor based self-adaptive compactor ride-on roller of claim 1, wherein, The sensor group (11) includes a pressure sensor, a displacement sensor and a temperature sensor.

6. Control method of a self-adapting compactor hand truck based on intelligent sensing, characterized in that, The method comprises the following steps: Step one, obtaining a plurality of sets of real-time data through the sensor group (11), using Kalman filtering algorithm to perform space-time alignment and noise suppression on the plurality of sets of real-time data, obtaining a processed data set, the data set comprising temperature data, pressure data and displacement data; Step two, obtaining a plurality of sets of historical pressure data to calculate the compaction degree influence coefficient, and obtaining historical operation data of the road roller, constructing a compaction degree evaluation model based on a convolutional neural network and combining the compaction degree influence coefficient; Step three, obtaining the data set, calculating the real-time compaction degree influence coefficient according to the real-time pressure data, inputting the compaction degree influence coefficient, the temperature data and the displacement data into the compaction degree evaluation model to output the current compaction degree value and the uniformity index; Step four, judging the current compaction state according to the preset compaction degree threshold and uniformity evaluation threshold, obtaining a compaction state judgment result, and adjusting the real-time operation parameters of the road roller in real time through the PID control algorithm, the real-time operation parameters including the output power of the power system, the amplitude of the rolling roller (3) and the moving speed of the road roller.

7. The control method of a smart sensor based self-adaptive compaction ride-on roller according to claim 6, characterized in that, The specific process of calculating the compaction degree influence coefficient is as follows: Obtain a data set, parse several groups of pressure data F and temperature data Ti from the data set, and calculate the compaction degree influence coefficient Ki according to the following formula after standardizing and de-dimensioning the force data F and the temperature data Ti: Wherein d and e are preset weight coefficients, is a preset pressure average value, i = 1, 2, 3, …, n, and the compaction degree influence coefficient is used to reflect the influence degree of the ground flatness to be processed and the environmental temperature on the compaction effect.

8. The control method of a smart sensor based self-adaptive compaction ride-on roller according to claim 6, characterized in that, The specific process of constructing the compaction degree evaluation model is as follows: S201, the historical operation data of the road roller includes compaction process parameters under different material types, temperature data and initial compaction degree, and a compaction degree influence coefficient, the historical operation data of the road roller is integrated as a training sample, and the training sample is split into a training set and a test set according to a proportion of 8:2; S202, a compaction degree evaluation model based on a convolutional neural network is constructed, a weight file is downloaded and loaded onto a corresponding network for initializing migration network parameters; S203, the last fully connected layer of the network is modified, the input is kept unchanged, the output is set to a compaction degree value and a uniformity index, the last layer is weight initialized, a gradient descent algorithm is used for learning, and fixed step attenuation is adopted to optimize training parameters, the entire network is retrained to obtain the compaction degree evaluation model; S204, small batches of training samples are randomly and repeatedly extracted from the training set during the training process, all training samples are extracted as a training period, the training is completed after a certain period of iteration, and the compaction degree evaluation model is obtained.

9. The control method of a smart sensor based self-adaptive compaction ride-on roller according to claim 6, characterized in that, The specific process of obtaining the compaction state judgment result is as follows: S301, the compaction degree influence coefficient, temperature data and displacement data are input into the compaction degree evaluation model to output a current compaction degree value Ai and a uniformity index Bi; S302, a preset compaction degree threshold Amin and a uniformity evaluation threshold Bmax are obtained, if the current compaction degree value Ai is greater than or equal to the compaction degree threshold Amin, and the uniformity index Bi is less than or equal to the uniformity evaluation threshold Bmax, the compaction state judgment result is excellent; if the current compaction degree value Ai is greater than or equal to the compaction degree threshold Amin, and the uniformity index Bi is greater than the uniformity evaluation threshold Bmax, the compaction state judgment result is good; if the current compaction degree value Ai is less than the compaction degree threshold Amin, and the uniformity index Bi is less than or equal to the uniformity evaluation threshold Bmax, the compaction state judgment result is good; if the current compaction degree value Ai is less than the compaction degree threshold Amin, and the uniformity index Bi is greater than the uniformity evaluation threshold Bmax, the compaction state judgment result is poor.

10. The control method of the smart sensor based self-adaptive compaction ride-on roller according to claim 6, characterized in that, The specific process of adjusting the real-time operation parameters of the road roller in real time through the PID control algorithm is as follows: S401, the BP neural network is combined with the PID controller, the input layer of the neural network is the multi-sensor data, the implicit layer approaches the optimal control law through self-learning, and the output layer directly adjusts the PID parameters; S402, the PID parameter control process is as follows: 1) Vibration frequency control: adjust the frequency through the vibration system, and the PID controller takes the error between the target frequency and the actual frequency as the input to dynamically adjust the opening degree of the hydraulic valve; 2) Traveling speed control: adopt an incremental PID algorithm (Δu(k)=P[e(k)-e(k-1)]+Ie(k)+D[e(k)-2e(k-1)+e(k-2)]) and the output is an engine speed adjustment signal, wherein e is the walking efficiency of the walking system and k is the efficiency influence coefficient; 3) Amplitude and steering control: Amplitude switching: according to the thickness of the layer, the corresponding amplitude is selected, and the pressure of the hydraulic system is adjusted by PID to realize rapid switching; Steering control: based on double GPS path planning information, the PID controller adjusts the steering angle, and the positive and negative transfer network algorithm is combined to ensure the trajectory tracking accuracy.

Images