CN110238810B

CN110238810B - Intelligent robot with improved image recognition technology based on deep learning

Info

Publication number: CN110238810B
Application number: CN201910513661.7A
Authority: CN
Inventors: 张培荣; 王虹杰; 方岳龙; 宋超; 商允良; 刘玉; 李乐政
Original assignee: Dongfang Electronics Co Ltd
Current assignee: Dongfang Electronics Co Ltd
Priority date: 2019-06-14
Filing date: 2019-06-14
Publication date: 2021-03-09
Anticipated expiration: 2039-06-14
Also published as: CN110238810A

Abstract

The invention discloses an intelligent robot with improved image recognition technology based on deep learning. The automobile comprises an automobile body, run through the first pivot that extends around the rotation is connected with on the automobile body right-hand member, first pivot front and back end stretches out the automobile body front and back end and the support balance plate of fixedly connected with front and back symmetry, first pivot front and back end is installed with the follow driving wheel with rotating, be equipped with the arc spout in the automobile body, link up in the inner wall around the arc spout be equipped with the endocentric arc of arc spout crosses the aqueduct, sliding connection has arc power piece in the arc spout, be equipped with the transmission chamber in the arc power piece, symmetry and central line are at the epaxial second pivot of same straight line around transmission chamber front and back end inner wall internal rotation is installed, the second pivot is kept away from arc power piece end runs through the aqueduct is crossed to the arc and is run through the support.

Description

Intelligent robot with improved image recognition technology based on deep learning

Technical Field

The invention relates to the technical field of power equipment inspection robots, in particular to an intelligent robot improved by an image recognition technology based on deep learning.

Background

In recent years, national power grids in China are gradually built and improved, and the total length and the transmission capacity of a power transmission line are increased year by year. The power distribution network mainly comprises power generation equipment, power transmission equipment, power transformation equipment, power distribution equipment and power utilization equipment, and the key point in the process of distribution network is equipment protection. In order to ensure the reliability of the transmission line and discover the damage and hidden danger of the line in time, the power department must regularly patrol the transmission line. The traditional power transmission line inspection is mainly a manual inspection mode, the mode is restricted by natural environment and inspection equipment, the cost is high, the efficiency is low, the missing inspection rate and the false inspection rate are high, and the effect is poor. At present, a power grid video monitoring system is taken as an important component of an intelligent power grid and widely applied to production management, power transmission and transformation line state monitoring, maintenance and the like of the power grid, but the video monitoring system also has certain limitations, large video data storage pressure, insufficient network advantage utilization and the like.

With the development of image processing and computer vision technologies, computer vision is used to identify power equipment, and analyzing the daily operating state of the power equipment plays an increasingly important role in power system detection, and it is particularly important to monitor the operating state of the power equipment in real time.

The method for extracting image features in the image recognition and classification method used in the current power equipment is a key step of image recognition, and when the image background is simple and the characteristics of the power equipment are prominent, the image color features, the image texture features, the image shape features and the like adopted in the traditional method can obtain a relatively ideal image recognition rate. However, the types of the power equipment are numerous and the colors are single, the environments of power transmission and transformation departments are complex, the identification accuracy of the power equipment is greatly influenced by the complex backgrounds in the images of the power equipment, and the appearance of the same power equipment is greatly changed under the influence of the change of shooting angles, distances, illumination and shadows, so that the traditional power equipment image feature extraction technology is difficult to meet the identification requirements and accuracy of the power equipment identification, in addition, the image identification classification method used in the current power equipment needs to manually label a large amount of sample data, is time-consuming and labor-consuming, has high economic cost, and is not ideal for the identification accuracy of the images, so that a graphic image identification method based on deep learning and used in the routing inspection of the power equipment is urgently needed, directly acts on the original data of the power equipment, automatically performs feature learning layer by layer, the global characteristics of the image are better described, intelligent analysis and identification of massive unstructured image data in the power equipment system are achieved, the accuracy of image identification of different power equipment is improved, early warning of power equipment accidents is completed, and the safety of the power equipment is guaranteed.

Disclosure of Invention

The invention aims to provide an intelligent robot improved by an image recognition technology based on deep learning, which fundamentally solves the problems that the existing image recognition method of power equipment needs to manually label a large amount of sample data, is time-consuming, labor-consuming, high in economic cost and too single in image feature extraction, cannot better describe the global features of an image and causes low recognition accuracy of the final image.

In order to achieve the purpose, the invention provides the following technical scheme: an intelligent robot with improved image recognition technology based on deep learning comprises a vehicle body, wherein a first rotating shaft extending forwards and backwards is rotatably connected to the right end of the vehicle body in a penetrating manner, the front end and the rear end of the first rotating shaft extend out of the front end and the rear end of the vehicle body and are fixedly connected with front and rear symmetrical supporting swinging plates, a driven wheel is rotatably mounted at the front end and the rear end of the first rotating shaft, an arc sliding groove is formed in the vehicle body, an arc transition groove concentric with the arc sliding groove is formed in the front inner wall and the rear inner wall of the arc sliding groove in a penetrating manner, an arc power block is slidably connected in the arc sliding groove, a transmission cavity is formed in the arc power block, a second rotating shaft which is symmetrical forwards and backwards and is provided with a central line on the same straight line is rotatably mounted in the inner wall of the front end and the rear end of the transmission, the second pivot runs through the terminal fixedly connected with action wheel of support wobble plate end, two second pivots be close to each other and end stretch into near each other the transmission intracavity and be equipped with separation and reunion link gear, be equipped with first bevel gear in the second pivot, the embedded motor that is equipped with front and back symmetry of transmission chamber left end inner wall, motor right-hand member output end fixedly connected with be used for with the second bevel gear that first bevel gear meshing is connected, with one side from the driving wheel with friction connection has the track between the action wheel, fixedly connected with is used for the restriction on the side of automobile body support the balanced stopper of wobble plate time of pivoted, the articulated hydraulic telescoping rod that is connected with on the side of automobile body, the hydraulic telescoping rod bottom with support the wobble plate and keep away from the side of automobile body is articulated to be connected. The top of automobile body is equipped with DSP focus balance detection case, and DSP focus balance detection case top is equipped with the detector fixture of full scope.

As a preferable technical scheme, the clutch linkage mechanism comprises a sleeve disc fixedly connected to the central end face of the second rotating shaft at the rear end and a rotating disc fixedly connected to the central end face of the second rotating shaft at the front end and sleeved with the sleeve disc, a circular groove for being sleeved with the rotating disc is formed in the front end face of the sleeve disc, a guide sliding groove is formed in the front end face of the rotating disc, the guide sliding groove penetrates through the side end face of the rotating disc, friction blocks which are bilaterally symmetrical and are in friction fit with the inner wall of the circular groove are connected in the guide sliding groove in a sliding mode, a sleeve is sleeved on the second rotating shaft at the front end, a connecting rod which is bilaterally symmetrical is hinged between the sleeve and the friction block, a shifting plate is rotatably connected to the sleeve, an opening is formed in the inner wall of the right end of the transmission cavity in a penetrating mode, and the shifting plate penetrates, the inner wall of the right end of the arc chute is provided with a yielding cavity, and the inner wall of the rear end of the yielding cavity is fixedly connected with an electric telescopic rod fixedly connected with the shifting plate.

According to the technical scheme, the center of the bottom of the inner cavity of the DSP gravity center balance detection box is fixedly connected with a supporting seat, the top of the supporting seat is fixedly connected with a supporting rod, the top of the supporting rod is fixedly connected with a hinged ball, the hinged ball is hinged with a balance disc, a plurality of X-ray generators which are distributed densely are arranged on the side face of the balance disc, and a light receiver which is parallel to the balance disc is arranged on the inner wall of the inner cavity of the DSP gravity center balance detection box. Through the effect of gravity balance, no matter when the automobile body is climbing the stair of any slope, it is parallel until the plane ray that light receiver and X ray generator sent is complete to adjustable DSP focus balance detection case to make the balanced detection of being convenient for of automobile body.

As preferred technical scheme, detector fixture is including digging circular slot on the balanced detection case top end face of DSP focus, the disc is installed to the circular slot internal rotation, disc bottom power is connected with servo motor, disc top center fixedly connected with fixed column, the articulated arm that is connected with the multisection is articulated in fixed column top, the articulated second hydraulic stretching pole that is connected with between every section of arm, the terminal fixedly connected with image acquisition ware of arm. Install the detector above that through image acquisition ware, then can realize the detection at the no dead angle of comprehensive maximum scope through the arm, promote detection effect and working category, the practicality improves greatly.

According to a preferable technical scheme, a hydraulic system is arranged in the vehicle body, the hydraulic system is connected with the hydraulic telescopic rod and the second hydraulic telescopic rod through oil pipes so as to control the hydraulic telescopic rod to stretch, a processor is arranged in the vehicle body, and the processor is electrically connected with the optical receiver and the hydraulic system. Through the effect of gravity balance, no matter the automobile body is when climbing the stair of any slope, the balanced detection case of adjustable DSP focus is parallel until light receiver and the plane ray that the X ray generator sent complete the joint, and at this in-process, the steerable hydraulic stem of hydraulic system is flexible to realize the balance of automobile body.

The invention has the beneficial effects that:

(1) the intelligent stair climbing device is simple in structure and convenient to operate, the crawler driving mechanism capable of changing the inclination angle through hydraulic driving is adopted, the angle adjustment of the crawler mechanism is provided with the DSP gravity center balance detection box according to the top, the gravity center can be readjusted to be balanced by combining digital signal processing, the balance is stable, the stair climbing is rapid, the types of stairs climbing are more, the detection range is wide, the image acquisition definition is high, the practicability is higher, the size is smaller, the movement is convenient, high-tech technology is combined, manufacturers favor products, the production and manufacturing probability is higher, the intelligent and stable stair climbing of any type can be realized, the problem that the cost is higher due to the limited types of stairs climbing is avoided, and the problem that the gravity center deviates and topples is solved. The method has high identification accuracy rate of the images of the power equipment, provides a new solution for intelligent analysis of massive unstructured images of the power equipment, better utilizes a large amount of unmarked data to improve the identification accuracy rate of the images, and has low economic cost.

Drawings

For ease of illustration, the invention is described in detail by the following specific examples and figures.

FIG. 1 is a schematic diagram of an internal overall structure of an intelligent robot improved by an image recognition technology based on deep learning according to the present invention;

FIG. 2 is a schematic cross-sectional view A-A of FIG. 1;

fig. 3 is a schematic view of the internal structure of the arc-shaped power block of the invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, are within the scope of the present invention

In the description of the present invention, it should be noted that the terms "upper", "lower", "inner", "outer", "top/bottom", and the like indicate orientations or positional relationships based on those shown in the drawings, and are only for convenience of description and simplification of description, but do not indicate or imply that the referred device or element must have a specific orientation, be constructed in a specific orientation, and be operated, and thus should not be construed as limiting the present invention. Furthermore, the terms "first" and "second" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.

In the description of the present invention, it should be noted that, unless otherwise explicitly specified or limited, the terms "mounted," "disposed," "sleeved/connected," "connected," and the like are to be construed broadly, e.g., "connected," which may be fixedly connected, detachably connected, or integrally connected; can be mechanically or electrically connected; they may be connected directly or indirectly through intervening media, or they may be interconnected between two elements. The specific meanings of the above terms in the present invention can be understood in specific cases to those skilled in the art.

Example 1:

referring to fig. 1-3, an intelligent robot based on the improved image recognition technology of deep learning comprises a vehicle body 1, a first rotating shaft 9 extending back and forth is rotatably connected to the right end of the vehicle body 1, the front and rear ends of the first rotating shaft 9 extend out of the front and rear ends of the vehicle body 1 and are fixedly connected with supporting swinging plates 6 symmetrical back and forth, a driven wheel 8 is rotatably mounted at the front and rear ends of the first rotating shaft 9, an arc chute 42 is arranged in the vehicle body 1, an arc transition groove 21 concentric with the arc chute 42 is arranged in the front and rear inner walls of the arc chute 42 in a penetrating manner, an arc power block 24 is slidably connected in the arc chute 42, a transmission cavity 27 is arranged in the arc power block 24, second rotating shafts 29 symmetrical back and forth are rotatably mounted on the front and rear inner walls of the transmission cavity 27 and have central lines on the same straight line, the ends of the second rotating shafts 29 far away from the arc power block 24 penetrate, the second pivot 29 runs through the terminal fixedly connected with action wheel 2 that supports balance 6 end, two second pivot 29 be close to each other and be close to the end and stretch into transmission chamber 27 and be equipped with separation and reunion link gear, be equipped with first bevel gear 28 on the second pivot 29, the embedded motor 25 that is equipped with front and back symmetry of transmission chamber 27 left end inner wall, motor 25 right-hand member output end terminal fixedly connected with is used for the second bevel gear 26 of being connected with first bevel gear 28 meshing, friction connection has track 7 between follow driving wheel 8 and the action wheel 2 with one side, fixedly connected with is used for the restriction to support when balance 6 of balance 5 stopper of balance 6 on the side of automobile body 1, articulated hydraulic telescoping rod 4 that is connected on the side of automobile body 1, articulated the connection of side that keeps away from automobile body 1 of hydraulic telescoping rod 4. The top of automobile body 1 is equipped with DSP focus balance detection case 10, and DSP focus balance detection case 10 top is equipped with the detector fixture of full scope.

The clutch linkage mechanism comprises a sleeve disc 34 fixedly connected on the end face of the central end of the second rotating shaft 29 at the rear end and a rotating disc 32 fixedly connected on the end face of the central end of the second rotating shaft 29 at the front end and used for being sleeved with the sleeve disc 34, a circular groove 35 used for being sleeved with the rotating disc 32 is formed in the end face of the front end of the sleeve disc 34, a guide sliding groove 33 is formed in the end face of the front end of the rotating disc 32, the guide sliding groove 33 penetrates through the end face of the side face of the rotating disc 32, friction blocks 41 which are bilaterally symmetrical and used for being in friction fit with the inner wall of the circular groove 35 are connected in the guide sliding groove 33 in a sliding mode, a sleeve 39 is sleeved on the second rotating shaft 29 at the front end, connecting rods 40 which are bilaterally symmetrical are hinged between the sleeve 39 and the friction blocks 41, a shifting plate 36 is connected on the sleeve 39 in a rotating mode, an opening 37 is, an electric telescopic rod fixedly connected with the poking plate 36 is fixedly connected to the inner wall of the rear end of the abdicating cavity 31.

The center of the bottom of the inner cavity of the DSP gravity center balance detection box 10 is fixedly connected with a supporting seat 19, the top of the supporting seat 19 is fixedly connected with a supporting rod 18, the top of the supporting rod 18 is fixedly connected with a hinged ball 3, the hinged ball 3 is hinged with a balance disc 16, the side surface of the balance disc 16 is provided with a plurality of densely distributed X-ray generators 17, and the inner wall of the inner cavity of the DSP gravity center balance detection box 10 is provided with a light receiver 11 parallel to the balance disc 16. Through the action of gravity balance, no matter when the vehicle body 1 climbs stairs with any gradient, the gravity center balance detection box 10 of the DSP can be adjusted to be parallel until the light receiver 11 is completely jointed with plane rays emitted by the X-ray generator 17, so that the vehicle body is balanced and convenient to detect.

The detector clamping mechanism comprises a circular groove formed in the top end face of the DSP gravity center balance detection box 10, a disc 20 is installed in the circular groove in a rotating mode, the bottom of the disc 20 is connected with a servo motor in a power mode, a fixed column 15 is fixedly connected to the center of the top of the disc 20, multiple sections of mechanical arms 14 connected in a hinged mode are hinged to the top of the fixed column 15, a second hydraulic telescopic rod 13 is hinged between every two sections of the mechanical arms 14, and the tail ends of the mechanical arms 14 are fixedly connected with an image collector 12. The detector is installed on the image acquisition device 12, and then the mechanical arm 14 can realize the detection without dead angles in the largest range, so that the detection effect and the working range are improved, and the practicability is greatly improved.

Be equipped with hydraulic system in the automobile body 1, hydraulic system passes through oil pipe connection with hydraulic telescoping rod 4 and second hydraulic telescoping rod 13, is convenient for control it is flexible, and automobile body 1 is inside to be equipped with the treater, treater and light receiver 11 and hydraulic system electrical connection. Through the action of gravity balance, no matter when the vehicle body 1 climbs stairs with any gradient, the gravity center balance detection box 10 of the DSP can be adjusted to be parallel until the light receiver 11 is completely jointed with plane rays emitted by the X-ray generator 17, and in the process, the hydraulic system can control the hydraulic rod 4 to stretch out and draw back to realize the balance of the vehicle body.

The invention relates to an intelligent robot improved by an image recognition technology for deep learning, which comprises the following specific steps:

s1, in an initial state, the vehicle body 1, the supporting swing plate 6, the caterpillar band 7, the DSP gravity center balance detection box 10 and the balance disc 16 are in a horizontal state, the supporting swing plate 6 is abutted to the limiting block 5, the mechanical arm 14 is in a storage state, the sleeve disc 34 and the rotary disc 32 are in a separation state, the friction block 41 is completely positioned in the guide chute 33, and the electric telescopic rod 30 is in the longest state.

S2: when the device needs to work, firstly, the images of the electric equipment to be collected are found, then the two motors 25 are started to drive the second bevel gear 26 to rotate, the first bevel gear 28 is meshed with the second bevel gear 26 to drive the second rotating shafts 29 on the front side and the rear side to rotate, so as to drive the driving wheel 2 to rotate, and finally, the crawler 2 is driven to rotate, so that the whole moving device is moved, when the backward turning is needed, the rear end motor 25 is closed, the front end motor is started to complete the backward turning, when the forward turning is needed, the front end motor 25 is closed, the rear end motor is started to complete the forward turning, when the moving device needs large power to move, the electric telescopic rod 30 is started to contract to pull the shifting plate 36 to move backward, when the sleeve 39 moves backward, the connecting rod 38 enables the two friction blocks 41 to move to two sides until the friction blocks abut against the circular groove 35, so as to connect the sleeve 34 and the rotating disc 32 into a whole, through the action of gravity balance, no matter when the vehicle body 1 climbs stairs with any gradient, the DSP gravity center balance detection box 10 can be adjusted to be parallel until the light receiver 11 is completely jointed with the plane rays emitted by the X-ray generator 17, so that the vehicle body is balanced and convenient to detect, and when the vehicle body reaches the side of the electric equipment to be collected, the mechanical arm 14 is started to align the image collector 12 to the electric equipment to be collected.

The above description is only an embodiment of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions that are not thought of through the inventive work should be included in the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope defined by the claims.

Claims

1. The utility model provides an intelligent robot of image recognition technology improvement based on degree of deep learning which characterized in that: it includes automobile body (1), run through rotation connection on automobile body (1) right-hand member and extend around first pivot (9), first pivot (9) front and back end is stretched out automobile body (1) front and back end and support wobble plate (6) of fixedly connected with front and back symmetry, first pivot (9) front and back end is installed with rotating from driving wheel (8), be equipped with arc spout (42) in automobile body (1), link up in arc spout (42) front and back inner wall be equipped with arc spout (42) concentric arc crosses aqueduct (21), sliding connection has arc power piece (24) in arc spout (42), be equipped with transmission chamber (27) in arc power piece (24), symmetry and central line second pivot (29) on same straight line around transmission chamber (27) front and back end inner wall is installed with rotating, second pivot (29) are kept away from arc power piece (24) end runs through arc crosses aqueduct (21) The tail end of the second rotating shaft (29) penetrating through the end of the supporting swing plate (6) is fixedly connected with a driving wheel (2), the two second rotating shafts (29) are close to each other and extend into the transmission cavity (27) and are provided with clutch linkage mechanisms, the second rotating shaft (29) is provided with a first bevel gear (28), the inner wall of the left end of the transmission cavity (27) is embedded with a front-back symmetrical motor (25), the tail end of the output end of the right end of the motor (25) is fixedly connected with a second bevel gear (26) which is meshed and connected with the first bevel gear (28), a crawler belt (7) is connected between the driven wheel (8) and the driving wheel (2) on the same side in a friction mode, and the side face of the vehicle body (1) is fixedly connected with a limiting block (5) used for limiting the rotation of the supporting swing plate (6) during balance, articulated on the side of automobile body (1) be connected with hydraulic telescoping rod (4), hydraulic telescoping rod (4) bottom with support balance board (6) and keep away from the side of automobile body (1) is articulated to be connected, the top of automobile body (1) is equipped with DSP focus balance detection case (10), and DSP focus balance detection case (10) top is equipped with the detector fixture of full scope.

2. The intelligent robot improved by the image recognition technology based on the deep learning according to the claim 1, characterized in that: the clutch linkage mechanism comprises a sleeve disc (34) fixedly connected to the end face of the center end of a second rotating shaft (29) at the rear end and a rotary disc (32) fixedly connected to the end face of the center end of a second rotating shaft (29) at the front end and sleeved with the sleeve disc (34), a circular groove (35) sleeved with the rotary disc (32) is formed in the end face of the front end of the sleeve disc (34), a guide sliding groove (33) is formed in the end face of the front end of the rotary disc (32), the guide sliding groove (33) penetrates through the end face of the side face of the rotary disc (32), a friction block (41) which is bilaterally symmetrical and is in friction fit with the inner wall of the circular groove (35) is connected in the guide sliding groove (33) in a sliding mode, a sleeve (39) is sleeved on the second rotating shaft (29) at the front end, and a connecting rod (40) which is bilaterally symmetrical is hinged and connected between the friction block (, the sleeve (39) is connected with a shifting plate (36) in a rotating mode, an opening (37) is formed in the inner wall of the right end of the transmission cavity (27) in a penetrating mode, the shifting plate (36) penetrates through the opening (37) to stretch out to the outside, a yielding cavity (31) is formed in the inner wall of the right end of the arc-shaped sliding groove (42), and the inner wall of the rear end of the yielding cavity (31) is fixedly connected with an electric telescopic rod fixedly connected with the shifting plate (36).

3. The intelligent robot improved by the image recognition technology based on the deep learning according to the claim 1, characterized in that: DSP focus balance detection case (10) inner chamber bottom center fixedly connected with supporting seat (19), supporting seat (19) top fixedly connected with bracing piece (18), bracing piece (18) top fixedly connected with hinge ball (3), the articulated balance disc (16) that is connected with on hinge ball (3), balance disc (16) side is equipped with a plurality of intensive X ray generators (17) that distribute, be equipped with on DSP focus balance detection case (10) inner chamber inner wall with the parallel light receiver (11) of balance disc (16).

4. The intelligent robot improved by the image recognition technology based on the deep learning according to the claim 3, characterized in that: detector fixture is including digging circular slot on DSP focus balance detection case (10) top end face, disc (20) are installed to the circular slot internal rotation ground, disc (20) bottom power is connected with servo motor, disc (20) top center fixedly connected with fixed column (15), articulated arm (14) that are connected with the multisection is articulated in fixed column (15) top, the articulated second hydraulic stretching pole (13) that is connected with between every section of arm (14), arm (14) terminal fixedly connected with image collector (12).

5. The intelligent robot improved by the image recognition technology based on the deep learning according to claim 4, wherein: be equipped with hydraulic system in automobile body (1), hydraulic system with hydraulic telescoping rod (4) with second hydraulic telescoping rod (13) pass through oil pipe connection, the inside treater that is equipped with of automobile body (1), the treater with light receiver (11) with hydraulic system electrical connection.