CN112013203B - A Pipe Network Detection System Based on DRNN Neural Network - Google Patents
A Pipe Network Detection System Based on DRNN Neural Network Download PDFInfo
- Publication number
- CN112013203B CN112013203B CN202010694962.7A CN202010694962A CN112013203B CN 112013203 B CN112013203 B CN 112013203B CN 202010694962 A CN202010694962 A CN 202010694962A CN 112013203 B CN112013203 B CN 112013203B
- Authority
- CN
- China
- Prior art keywords
- neural network
- standard
- displacement
- pipe
- regulator
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F16—ENGINEERING ELEMENTS AND UNITS; GENERAL MEASURES FOR PRODUCING AND MAINTAINING EFFECTIVE FUNCTIONING OF MACHINES OR INSTALLATIONS; THERMAL INSULATION IN GENERAL
- F16L—PIPES; JOINTS OR FITTINGS FOR PIPES; SUPPORTS FOR PIPES, CABLES OR PROTECTIVE TUBING; MEANS FOR THERMAL INSULATION IN GENERAL
- F16L55/00—Devices or appurtenances for use in, or in connection with, pipes or pipe systems
- F16L55/26—Pigs or moles, i.e. devices movable in a pipe or conduit with or without self-contained propulsion means
- F16L55/28—Constructional aspects
- F16L55/30—Constructional aspects of the propulsion means, e.g. towed by cables
- F16L55/32—Constructional aspects of the propulsion means, e.g. towed by cables being self-contained
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F16—ENGINEERING ELEMENTS AND UNITS; GENERAL MEASURES FOR PRODUCING AND MAINTAINING EFFECTIVE FUNCTIONING OF MACHINES OR INSTALLATIONS; THERMAL INSULATION IN GENERAL
- F16L—PIPES; JOINTS OR FITTINGS FOR PIPES; SUPPORTS FOR PIPES, CABLES OR PROTECTIVE TUBING; MEANS FOR THERMAL INSULATION IN GENERAL
- F16L55/00—Devices or appurtenances for use in, or in connection with, pipes or pipe systems
- F16L55/26—Pigs or moles, i.e. devices movable in a pipe or conduit with or without self-contained propulsion means
- F16L55/28—Constructional aspects
- F16L55/40—Constructional aspects of the body
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F16—ENGINEERING ELEMENTS AND UNITS; GENERAL MEASURES FOR PRODUCING AND MAINTAINING EFFECTIVE FUNCTIONING OF MACHINES OR INSTALLATIONS; THERMAL INSULATION IN GENERAL
- F16L—PIPES; JOINTS OR FITTINGS FOR PIPES; SUPPORTS FOR PIPES, CABLES OR PROTECTIVE TUBING; MEANS FOR THERMAL INSULATION IN GENERAL
- F16L2101/00—Uses or applications of pigs or moles
- F16L2101/30—Inspecting, measuring or testing
Landscapes
- Engineering & Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- Chemical & Material Sciences (AREA)
- Combustion & Propulsion (AREA)
- Mechanical Engineering (AREA)
- Feedback Control In General (AREA)
Abstract
本发明涉及液压管应用领域，公开一种基于DRNN神经网络管网检测系统，其特征在于：主体管的两端分别连接有能够与待检测管道的检测部两侧螺纹连接的活动连接头(3)和固定连接头(4)；输送管与主体管的中部连通，标准器通过输送机构输送至输送管(2)内，且标准器的探头朝下位于主体管与输送管的衔接处，用于控制标准器位移的标准器位移控制系统包括标准器位移调节平台和MSP430单片机中的智能控制器。本发明能平稳地将标准器传送到检定位置，能有效避免标准器的探头触碰到输送管内壁引起参数变化，能够灵活调节标准器放置位置，减少人为操作引的标准器波动导致的测量结果不准确的情况。
The invention relates to the application field of hydraulic pipes, and discloses a pipe network detection system based on DRNN neural network, which is characterized in that two ends of a main pipe are respectively connected with movable connectors (3 ) and the fixed connector (4); the conveying pipe is communicated with the middle of the main body pipe, the standard device is conveyed into the conveying pipe (2) through the conveying mechanism, and the probe of the standard device is located at the junction of the main body pipe and the conveying pipe. The standard displacement control system for controlling standard displacement includes standard displacement adjustment platform and intelligent controller in MSP430 single-chip microcomputer. The invention can smoothly transfer the standard device to the verification position, can effectively avoid parameter changes caused by the probe of the standard device touching the inner wall of the conveying pipe, can flexibly adjust the placement position of the standard device, and reduce the measurement results caused by the fluctuation of the standard device caused by human operation. inaccurate situation.
Description
Technical Field
The invention relates to the field of application of hydraulic pipes, in particular to a pipe network detection system based on a DRNN neural network.
Background
In present pipeline industry, want the fluid pressure degree in the measurement pipeline, can reserve the measurement section pipeline at the section of measuring of the needs of pipeline when the installation pipeline usually, when needing to measure, take off the measurement section pipeline, detecting system detects at this position in the installation, under the normal condition, calibrator among the detecting system needs artifical a calibrator pipeline through detecting system to stretch into the pipeline that awaits measuring and examines the position and detect, when artifical with the calibrator insertion calibrator conveyer pipe, hand shake can take place for inevitable, make the bottom probe laminating of calibrator at the inner wall of calibrator conveyer pipe, perhaps relative shake influences the measurement accuracy between the bottom probe of calibrator and the rivers that await measuring in the pipeline.
At present, in the measurement and determination activities or experiments, the requirement of personnel on the experimental results is higher, but in the detection system in the market at present, the placement of the standard device under manual operation can cause a lot of irregular problems, thereby causing the problems of inaccurate measurement or inconsistency of parameters and actual values and the like.
Disclosure of Invention
The purpose of the invention is as follows: in order to solve the problems, the technical problem to be solved by the invention is to provide a pipe network detection system based on a DRNN neural network, which can stably transmit a standard device to a detection position, effectively avoid parameter change caused by the fact that a detection probe of the standard device touches the inner wall of a conveying pipe in the transmission process, conveniently adjust the placement position of the standard device, and reduce the situation that the measurement result is inaccurate due to the fluctuation of the standard device caused by manual operation; aiming at the reasons of nonlinearity, large hysteresis and large inertia of the displacement adjustment of the standard device, an intelligent standard device displacement control system is designed in a pipeline detection system so as to meet the requirement of displacement accuracy of the standard device in pipeline detection. The invention effectively overcomes the defect of inaccurate displacement adjustment of the existing standard, and improves the quick response, control precision and robustness of the displacement of the standard by designing an intelligent standard displacement control system in a pipeline detection system, so that the displacement of the standard reaches the set value of the system quickly, thereby meeting the requirement of the pipeline detection system on the precision control of the displacement of the standard.
The technical scheme is as follows: the invention provides a pipe network detection system based on a DRNN neural network, which comprises a main pipe, a conveying mechanism arranged at the pipe orifice of the conveying pipe and an intelligent standard device displacement control system used for controlling the displacement of a standard device, wherein two ends of the main pipe are respectively connected with a movable connector and a fixed connector which can be in threaded connection with two sides of a detection part of a pipeline to be detected; the conveying pipe is communicated with the middle part of the main body pipe, a standard device fixed with a standard device displacement sensor is conveyed into the conveying pipe through the conveying mechanism, and a probe of the standard device faces downwards and is positioned at the joint of the main body pipe and the conveying pipe; the intelligent standard displacement control system comprises a standard displacement adjusting platform and an intelligent standard displacement controller in an MSP430 single chip microcomputer, wherein the standard displacement adjusting platform consists of the MSP430 single chip microcomputer, an L298 motor driving circuit, the conveying mechanism and the standard displacement sensor; the intelligent etalon displacement controller in the MSP430 single chip microcomputer is composed of a parameter self-adjusting fuzzy regulator, a DRNN neural network regulator, a NARX neural network fusion regulator, a time sequence DRNN neural network predictor, a time sequence least square support vector machine (LS-SVM) predictor and a wavelet neural network fusion device.
Preferably, in the intelligent etalon displacement controller in the MSP430 single chip microcomputer, the parameter self-adjusting fuzzy regulator and the DRNN neural network regulator are connected in parallel, the outputs of the parameter self-adjusting fuzzy regulator and the DRNN neural network regulator are used as the inputs of the NARX neural network convergence controller, the output of the NARX neural network convergence controller is used as the input of the L298 motor driving circuit, and the output of the L298 motor driving circuit is used as the input of the driving motor in the conveying mechanism; the output of the etalon displacement sensor is respectively used as the input of the time series DRNN neural network predictor and the time series least square support vector machine (LS-SVM) predictor, the output of the time series DRNN neural network predictor and the time series least square support vector machine (LS-SVM) predictor is respectively used as the input of the wavelet neural network fusion device, the output value of the wavelet neural network fusion device is used as the etalon displacement feedback value of the intelligent etalon displacement control system, and the error change rate of the etalon displacement given value of the intelligent etalon displacement control system and the output value of the wavelet neural network fusion device are respectively used as the input of the parameter self-adjusting fuzzy regulator and the DRNN neural network regulator. An intelligent standard device displacement controller in an MSP430 singlechip realizes intelligent accurate control and regulation on standard device displacement, an NARX neural network fusion controller realizes fusion of a parameter self-regulation fuzzy regulator output value and a DRNN neural network regulator output value and the next prediction control on standard device displacement, and a wavelet neural network fusion device realizes fusion of a time series DRNN neural network predictor output value and a time series least square support vector machine (LS-SVM) predictor output value and the next accurate prediction on standard device displacement; the output of the NARX neural network fusion controller adjusts the control value of the displacement of the standard device, and the parameter self-adjusting fuzzy adjuster, the DRNN neural network adjuster and the NARX neural network fusion controller form composite control on the displacement of the standard device; the time series DRNN neural network predictor and the time series least square support vector machine (LS-SVM) predictor respectively predict the displacement of the standard device, the wavelet neural network fusion device fuses output values of the time series DRNN neural network predictor and the time series least square support vector machine (LS-SVM) predictor, the time series DRNN neural network predictor, the time series least square support vector machine (LS-SVM) predictor and the wavelet neural network fusion device form composite prediction of the displacement of the standard device, and the intelligent standard device displacement control system improves robustness, rapidity and accuracy of standard device displacement control.
Preferably, in the etalon displacement adjusting platform, the output of the NARX neural network fusion controller of the intelligent etalon displacement controller in the MSP430 single chip microcomputer serves as the input of an L298 motor driving circuit, the L298 motor driving circuit serves as the input of the driving motor in the conveying mechanism, the conveying mechanism drives the etalon to move, the etalon displacement sensor measures the amount of etalon movement, and the outputs of the etalon displacement sensor serve as the input of the time series DRNN neural network predictor and the time series least squares support vector machine (LS-SVM) predictor of the intelligent etalon displacement controller in the MSP430 single chip microcomputer respectively.
Preferably, the conveying mechanism comprises a shell, a driving motor, a driving tooth, a driven tooth, a driving annular conveying belt, a driven annular conveying belt, and a plurality of driving conveying belt positioning shafts and driven conveying belt positioning shafts; the shell is fixed at the pipe orifice of the conveying pipe, an output shaft of the driving motor is fixedly connected with a driving shaft of the driving tooth, the driving shaft and a driven shaft of the driven tooth are parallel to each other and are rotatably connected in the shell, and the driving tooth is meshed with the driven tooth; the driving transmission belt positioning shafts and the driven transmission belt positioning shafts are sequentially and parallelly installed up and down along two sides of the inner wall of the conveying pipe respectively; one end of the driving annular conveyor belt is sleeved on the driving shaft, and the middle part and the tail end of the driving annular conveyor belt are respectively sleeved on the positioning shafts of the driving conveyor belts; one end of the driven transmission belt is sleeved on the driven shaft, and the middle part and the tail end of the driven transmission belt are respectively sleeved on the driven transmission belt positioning shafts. When the standard device is required to be conveyed into the conveying pipe, the detection probe of the standard device is downwards plugged between the driving annular conveying belt and the driven annular conveying belt from the standard device inlet on the shell, when the driving gear is driven to rotate clockwise, the driving shaft rotates clockwise to drive the driving annular conveying belt to rotate clockwise, and then the positioning shafts of the driving conveying belts rotate clockwise; the driven teeth are meshed with the driving teeth, so that the driven teeth rotate anticlockwise, the driven shaft rotates anticlockwise to drive the driven annular transmission belts to rotate anticlockwise, and the positioning shafts of the driven transmission belts rotate anticlockwise; the driving transmission belt and the driven transmission belt rotate in opposite directions, so that the standard device can be conveyed from the upper part to the lower part of the conveying pipe; if the conveying position is too low, the standard device can be moved upwards to adjust the position by driving the driving teeth in the reverse direction; the conveying mechanism can enable the standard device to be stably conveyed to and positioned at the joint of the conveying pipe and the main pipe for measurement, reduces standard device fluctuation caused by manual operation, and can effectively avoid parameter change caused by the fact that a detection probe of the standard device touches the inner wall of the conveying pipe in the conveying process.
Preferably, the driving transmission belt positioning shafts, the driven transmission belt positioning shafts, the driving shaft and the driven shaft are all arranged in parallel.
Preferably, the distance between the driving endless transmission belt and the driven endless transmission belt is slightly smaller than the minimum width of the standard to be transmitted. The design can guarantee that the standard device can be clamped between the driving annular transmission belt and the driven annular transmission belt, and stable transmission is achieved.
Further, the dust cover is installed at the top of casing, the dust cover with the open-top of casing passes through spout sliding connection. When the dustproof cover is opened, the standard device can be thrown into a position between the driving annular transmission belt and the driven annular transmission belt from the opening at the upper part of the shell, and then the standard device is transmitted; when the detection is not carried out, the dustproof cover is closed to prevent dust from entering the conveying pipe.
The working principle is as follows: when liquid hydraulic pressure in a pipeline to be detected needs to be detected, firstly, the pipeline reserved at a detection part is detached, a fixed connector of the pipeline network detection system is in threaded connection with one side of a detection part of the pipeline to be detected, then a movable connector is in threaded connection with the other side of the detection part of the pipeline to be detected, then a standard is placed in a conveying mechanism, the conveying mechanism is controlled by a standard displacement control system to convey a detection probe of the standard downwards to a joint of a main pipe and a conveying pipe, and conveying is stopped, wherein the joint is a detection point. In the whole detection process, the standard device cannot shake or tremble, and the hydraulic pressure in the pipeline to be detected can be measured more accurately.
Has the advantages that: compared with the prior art, the invention has the following obvious advantages:
the input of the NARX neural network fusion regulator adopted by the invention comprises the input and output historical feedback of the parameter self-adjusting fuzzy regulator and the DRNN neural network regulator for a period of time, the feedback input can be considered to comprise the output historical information of the parameter self-adjusting fuzzy regulator and the DRNN neural network regulator for a period of time to participate in the control of the displacement of the standard device, the NARX neural network fusion regulator obtains good effect for a proper feedback time length, and the NARX neural network fusion regulator mode of the invention provides an effective standard device prediction control method.
The NARX neural network fusion regulator adopted by the invention is a dynamic neural network model which can effectively carry out prediction control on the nonlinear and non-stationary time sequence of the displacement of the standard device, and can improve the prediction control precision of the time sequence of the standard device under the condition of reducing the non-stationary time sequence. Compared with the traditional prediction control model method, the method has the advantages of good effect of processing the non-stationary time sequence, high calculation speed and high accuracy. Through the actual comparison of experimental data of a non-stationary standard, the method verifies the feasibility of the NARX neural network fusion regulator on the prediction control of the time series of the standard. Meanwhile, the experimental result also proves that the NARX neural network fusion regulator has more excellent performance in the non-stationary time series predictive control than the traditional predictive control model.
Thirdly, the invention utilizes the NARX neural network fusion regulator to establish the standard device predictive controller, because the dynamic recursive network of the model of the NARX neural network fusion regulator is established by introducing the delay module and the output feedback, the output of the parameter self-adjusting fuzzy regulator and the DRNN neural network regulator is taken as the input and the output vector delay feedback of the NARX neural network fusion regulator is introduced into the network training, the input of the NARX neural network fusion regulator not only comprises the output data of the original parameter self-adjusting fuzzy regulator and the DRNN neural network regulator, but also comprises the output data of the trained NARX neural network fusion regulator, and the generalization capability of the NARX neural network fusion regulator is improved, so that the NARX neural network fusion regulator has better prediction control precision and self-adaptive capability in nonlinear standard device time sequence prediction control compared with the traditional static neural network control.
The invention adopts a time series least square support vector machine (LS-SVM) predictor to predict the displacement of the standard device, utilizes the small sample of the displacement change of the standard device and the excellent predictive performance of nonlinear data to predict the displacement of the standard device, and predicts the displacement data of the standard device with high precision through the displacement time series variable of the original standard device.
And fifthly, the invention adopts a parameter self-adjusting fuzzy controller as a controller of the displacement of the standard device, the parameter of the fuzzy controller carries out self-adjustment according to the error e and the error change rate e' pair of the set value of the displacement of the standard device and the displacement predicted value of the standard device, and the fuzzy controller has strong robustness, high adaptability and high speed on the displacement control of the standard device.
And sixthly, the time series DRNN neural network predictor is a dynamic regression neural network with feedback and the capability of adapting to time-varying characteristics, the network can more directly and vividly reflect the dynamic change performance of the displacement of the standard device and can more accurately predict the actual value of the displacement of the standard device, the time series DRNN neural network standard device displacement predictor is of a 3-layer network structure of 10-21-1, the hidden layer of the time series DRNN neural network predictor is a regression layer, and the output layer of the time series DRNN neural network predictor is a standard device displacement prediction value.
The displacement of the standard device has the characteristics of nonlinearity, time lag and time variation as a controlled object, the parameter self-adjusting fuzzy controller, the DRNN neural network regulator and the NARX neural network fusion controller are combined with the advantages of the neural network and the fuzzy controller, the parameter self-adjusting fuzzy controller automatically adjusts the parameters of the fuzzy controller according to the error and the error variation of the displacement set value of the system standard device and the displacement predicted value of the standard device, and the intelligent composite controller has strong self-adaptability, improves the response speed of the system, can adapt to the influence of a plurality of disturbance factors and has good robustness.
The invention relates to the technology of neural network control, fuzzy control, composite control and composite prediction control, designs an intelligent standard displacement controller, and the control system has the intelligent standard displacement controller with good dynamic performance, high steady-state precision and stronger robustness, overcomes the defects of poor regulation quality and weak anti-interference performance of a pure PID control on large inertia and large delay objects, and has stronger dynamic tracking performance, anti-interference capability and good dynamic and static performance indexes when being used for controlling the displacement of the standard. Compared with the original conventional control, the control system has the advantages that the control quality, the response speed and the stability are obviously improved, the control precision of the displacement of the standard device is high, the anti-interference capability is high, the stability is good, and the application and popularization values are good.
Drawings
Fig. 1 and 2 are schematic perspective structural diagrams of a DRNN-based neural network pipe network detection system in embodiments 1 and 2;
fig. 3 and 4 are sectional perspective views of a part of the structure of the DRNN-based neural network pipe network detection system in embodiment 1;
FIG. 5 is a schematic diagram of an intelligent etalon displacement control system.
Detailed Description
The invention is further described with reference to the following figures and detailed description.
The embodiment provides a pipe network detection system based on a DRNN neural network, as shown in fig. 1 to 4, the pipe network detection system mainly comprises a main pipe 1, a conveying pipe 2, a conveying mechanism arranged at the pipe orifice of the conveying pipe 2 and a standard device displacement control system used for controlling the displacement of a standard device, wherein two ends of the main pipe 1 are respectively connected with a movable connector 3 (the movable connector 3 is a 304 stainless steel inner and outer wire straight-through connector with the model of ShBd sold in the market, and is connected to one end of the main pipe 1 after being directly purchased back) and a fixed connector 4 (the fixed connector 4 is a nut moving conversion loose-joint inner wire connector with the model of 304 stainless steel pipe sold in the market, and is connected to the other end of the main pipe 1 after being directly purchased back) which are in threaded connection with two sides of a detection part of a pipeline to be detected; the bottom of conveyer pipe 2 communicates with the middle part of main part pipe 1, and the etalon that is fixed with etalon displacement sensor passes through conveying mechanism and carries to conveyer pipe 2 in, and the probe of etalon is located the junction department of main part pipe 1 and conveyer pipe 2 downwards.
The conveying mechanism mainly comprises a shell 5, a driving motor 16, a driving tooth 6, a driven tooth 7, a driving annular conveying belt 8, a driven annular conveying belt 9, a plurality of driving conveying belt positioning shafts 10 and driven conveying belt positioning shafts 11; the shell 5 is fixed at the pipe orifice of the conveying pipe 2, the driving shafts 12 of the driving teeth 6 and the driven shafts 13 of the driven teeth 7 are parallel to each other and are rotatably connected in the shell 5, the driving teeth 6 are meshed with the driven teeth 7, and the driving motor 16 is fixedly connected with the driving shafts 12 of the driving teeth 6 and is positioned outside the shell 5; each driving transmission belt positioning shaft 10 and each driven transmission belt positioning shaft 11 are respectively arranged in parallel from top to bottom along two sides of the inner wall of the conveying pipe 2; one end of the driving annular conveyor belt 8 is sleeved on the driving shaft 12, and the middle part and the tail end of the driving annular conveyor belt are respectively sleeved on the driving conveyor belt positioning shafts 10; one end of the driven-shaped conveying belt 9 is sleeved on the driven shaft 13, and the middle part and the tail end of the driven-shaped conveying belt are respectively sleeved on the driven conveying belt positioning shafts 11. The distance between the driving endless transmission belt 8 and the driven endless transmission belt 9 is slightly smaller than the minimum width of the standard to be transmitted. The top of the housing 5 is provided with a dust cover 18, and the dust cover 18 is slidably connected with the top opening of the housing 5 through a sliding groove 19.
As shown in fig. 5, the intelligent etalon displacement control system comprises a etalon displacement adjusting platform and an intelligent etalon displacement controller in an MSP430 single chip microcomputer, wherein the etalon displacement adjusting platform consists of the MSP430 single chip microcomputer, an L298 motor driving circuit, the conveying mechanism and the etalon displacement sensor; the intelligent etalon displacement controller in the MSP430 single chip microcomputer is composed of a parameter self-adjusting fuzzy regulator, a DRNN neural network regulator, a NARX neural network fusion regulator, a time sequence DRNN neural network predictor, a time sequence least square support vector machine (LS-SVM) predictor and a wavelet neural network fusion device.
In the above standard displacement adjusting platform, the output of the NARX neural network fusion controller of the intelligent standard displacement controller in the MSP430 single chip microcomputer is used as the input of the L298 motor driving circuit, the L298 motor driving circuit is used as the input of the driving motor 16 in the conveying mechanism, the conveying mechanism drives the standard to move, the standard displacement sensor measures the movement amount of the standard, and the output of the standard displacement sensor is respectively used as the input of the time series DRNN neural network predictor and the time series least square support vector machine (LS-SVM) predictor of the intelligent standard displacement controller in the MSP430 single chip microcomputer.
In the intelligent standard displacement controller in the MSP430 single chip microcomputer, a parameter self-adjusting fuzzy regulator and a DRNN neural network regulator are connected in parallel, the output of the parameter self-adjusting fuzzy regulator and the output of the DRNN neural network regulator are used as the input of an NARX neural network fusion controller, the output of the NARX neural network fusion controller is used as the input of an L298 motor driving circuit, and the output of the L298 motor driving circuit is used as the input of a driving motor 16 in a conveying mechanism; the output of the standard device displacement sensor is respectively used as the input of a time series DRNN neural network predictor and a time series least square support vector machine (LS-SVM) predictor, the output of the time series DRNN neural network predictor and the time series least square support vector machine (LS-SVM) predictor is respectively used as the input of a wavelet neural network fusion device, the output value of the wavelet neural network fusion device is used as the standard device displacement feedback value of an intelligent standard device displacement control system, and the error change rate of the standard device displacement set value of the intelligent standard device displacement control system and the output value of the wavelet neural network fusion device are respectively used as the input of a parameter self-adjusting fuzzy regulator and a DRNN neural network regulator.
The working principle of the DRNN-based neural network pipe network detection system in the embodiment is as follows:
when the standard device is required to be transmitted into the conveying pipe 2 for measurement, the dust cover 18 is firstly opened along the sliding groove 19, then the detection probe of the standard device is downwards plugged between the driving annular conveying belt 8 and the driven annular conveying belt 9 from the top opening of the shell 5, the driving motor 16 is controlled to operate through the intelligent controller in the MSP430 single chip microcomputer, and when the driving gear 6 is driven by the driving motor 16 to rotate clockwise, the driving shaft 12 rotates clockwise to drive the driving annular conveying belt 8 to rotate clockwise, and then the positioning shafts 10 of the driving conveying belts rotate clockwise; the driven teeth 7 are meshed with the driving teeth 6, so that the driven teeth 7 rotate anticlockwise, the driven shaft 13 rotates anticlockwise to drive the driven annular transmission belt 9 to rotate anticlockwise, and the driven transmission belt positioning shafts 11 rotate anticlockwise; the driving annular conveying belt 8 and the driven annular conveying belt 9 rotate in opposite directions, so that the standard device can be conveyed to the joint of the conveying pipe 2 and the main pipe 1 from the pipe orifice of the conveying pipe 2; during the conveying process, the standard displacement sensor measures the movement amount of the standard, and when the movement amount measured by the standard displacement sensor reaches a preset value, the intelligent controller controls the driving motor 16 to stop running. After the detection is finished, the driving motor 16 is controlled to reversely rotate through an intelligent controller in the MSP430 single chip microcomputer, and the driving tooth 6 is reversely driven to move the standard device upwards to the opening at the top of the shell 5 to be taken out.
The overall function of the intelligent standard displacement controller is designed as follows:
(1) parameter self-adjusting fuzzy controller design
The invention relates to a patent parameter self-adjusting fuzzy controller as a prediction controller of standard device displacement, which is connected with a DRNN neural network regulator in parallel to realize the composite control of the standard device displacement and consists of two parts of fuzzy control and integral action which are connected in parallel. Its fuzzy control rule is uf＝k0X f (e, e'), wherein: u. offAdjusting the output of the fuzzy controller for the parameter; k is a radical of0Is the output coefficient; f (e, e ') is an adaptive control rule function, and the fuzzy control rule is f (e, e ') - α × e + (1- α) e ', wherein: alpha is an adaptive correction factor, and alpha is more than or equal to 0 and less than or equal to 1; the magnitude of alpha reflects the degree of influence of the error e and the error change rate e' of the set value of the standard displacement and the composite predicted value of the standard displacement on the output of the parameter self-adjusting fuzzy controller. By analyzing the functions of e and e 'in different stages of the displacement control of the standard device, the influence of the e and e' on the parameter self-adjusting fuzzy controller in different control stages is different. In the early stage, if e and e' of the etalon are of opposite sign, the initial error is relatively large, which isA larger value of alpha should be chosen to eliminate the presence of etalon displacement errors as quickly as possible. Therefore, the weight of the error in the parameter self-adjusting fuzzy control rule should be increased; in the middle stage, the displacement error of the standard is reduced, the rising speed of the intelligent standard displacement control system is accelerated, the control action on the displacement error change of the standard is highlighted for reducing the overshoot of the intelligent standard displacement control system, and a smaller alpha value is selected; when the etalon response approaches the desired value, both may be weighted equally, since the etalon displacement error and its variation are now small. In the practical realization process, the selection of the alpha value is obtained by a table look-up program, the input fuzzy variable of the parameter self-adjusting fuzzy controller of the patent is the error e and the error change rate e' of the set value of the standard displacement and the composite predicted value of the standard displacement, the output quantity is the prediction control quantity of the parameter self-adjusting fuzzy controller, and the basic discourse domain of the input fuzzy variable is [ -2, 2]The quantization discourse field is [ -3, 3 [)]Therefore, the quantization factor k1 is 1.5; the table look-up based on the corresponding error is as follows:
wherein alpha is0,α1,α2,α3∈[0,1]In general, alpha0＜α1＜α2＜α3Therefore, different requirements of the displacement control system of the standard device on the correction factor under different working conditions can be met.
(2) DRNN neural network regulator design
The DRNN neural network regulator is a dynamic regression neural network with feedback and the ability of adapting to time-varying characteristics, the network can more directly and vividly reflect the dynamic change performance of the displacement of the standard device and can more accurately regulate the displacement of the standard device, and the hidden layer of the 3-layer network structure of each DRNN network 2-7-1 is a regression layer. In the DRNN neural network model of the present invention, let I ═ I1(t),I2(t),…,In(t)]Inputting a vector for the network, wherein Ii(t) error and error rate of change of etalon displacement is DRNNInputting the ith neuron of the input layer through the network regulator at time t, and outputting the jth neuron of the regression layer as Xj(t)，Sj(t) is the sum of the j-th regression neuron input, f (-) is a function of S, then O (t) is the output of the DRNN neural network modulator as the etalon displacement control. The output of the DRNN network regulator is:
(3) NARX neural network fusion controller design
The inputs of the NARX neural network fusion controller are respectively the outputs of the DRNN neural network regulator and the parameter self-adjusting fuzzy controller, and the NARX neural network fusion controller realizes the fusion of the output control quantity of the DRNN neural network regulator and the parameter self-adjusting fuzzy controller, so that the accuracy of the displacement control quantity of the standard device is further improved. The NARX neural network fusion controller (Nonlinear Auto-Regression with External input neural network) is a dynamic feedforward neural network, the NARX neural network fusion controller is a Nonlinear autoregressive network with the input of a DRNN neural network regulator and a parameter self-adjusting fuzzy controller, the NARX neural network fusion controller has the dynamic characteristic of multi-step time delay and is connected with a plurality of layers of closed networks through feedback, the NARX neural network fusion controller is a dynamic neural network which is most widely applied in a Nonlinear dynamic system, and the performance of the NARX neural network fusion controller is generally superior to that of a full-Regression neural network. A typical NARX neural network fusion controller mainly comprises an input layer, a hidden layer, an output layer and input and output delays, wherein before application, the delay order and the hidden layer neuron number of the input and the output are generally determined in advance, and the current output of the NARX neural network fusion controller not only depends on the past output y (t-n), but also depends on the current input vector DRNN neural network regulator and the delay order of a parameter self-adjusting fuzzy controller. The NARX neural network fusion controller comprises an input layer, an output layer, a hidden layer and a time-extension layer. The DRNN neural network regulator and the parameter self-adjusting fuzzy controller are transmitted to the hidden layer through the time delay layer, the hidden layer processes signals output by the DRNN neural network regulator and the parameter self-adjusting fuzzy controller and then transmits the processed signals to the output layer, the output layer performs linear weighting on the output signals of the hidden layer to obtain final output signals of the NARX neural network fusion controller, and the time delay layer delays signals fed back by the network and signals output by the input layer and then transmits the delayed signals to the hidden layer. The NARX neural network fusion controller has the characteristics of non-linear mapping capability, good robustness, adaptability and the like. x (t) represents the external inputs to the NARX neural network, i.e., the DRNN neural network regulators and parameter self-adjusting fuzzy controller output values; m represents the delay order of the external input; y (t) is the output of the NARX neural network, namely the output control quantity of the NARX neural network fusion controller in the next time period; n is the output delay order; s is the number of hidden layer neurons; the output of the jth implicit element can thus be found as:
in the above formula, wjiAs a connection weight between the ith input and the jth implicit neuron, bjIs the bias value of the jth implicit neuron, the output y (t +1) of the NARX neural network convergence controller has the value:
y(t+1)＝f[y(t),y(t-1),…,y(t-n),x(t),x(t-1),…,x(t-m+1)；W] (4)
(4) time series least squares support vector machine (LS-SVM) predictor design
The predictor of the time series least square support vector machine (LS-SVM) has stronger generalization capability and global capability, overcomes the defects of poor generalization capability, overfitting, easy falling into local optimum and the like of other machine learning methods, is an extension to a standard support vector machine, adopts a sum-of-squares error loss function to replace an insensitive loss function of the standard support vector machine, and simultaneously realizes the conversion of inequality constraint in a standard SVM algorithm into equal constraint. Therefore, a time series least square support vector machine (LS-SVM) predictor reduces the quadratic programming problem into solving a linear equation set, and obviously reduces the complexity of solvingThe calculation speed is improved. The method comprises the steps of setting a predictive value data training sample set D { (x) of a least square support vector machine (LS-SVM)i,yi)|i＝1,2,…,n}，xiAnd yiInput and output sample data, respectively, and n is the number of samples, which can map the input samples from the original space to the high-dimensional feature space. Introducing a Lagrange equation, converting an optimization problem with constraint conditions into an optimization problem without constraint conditions, and obtaining a linear regression equation of a time series least square support vector machine (LS-SVM) predictor as follows:
in the solving process, in order to avoid solving a complex nonlinear mapping function, a Radial Basis Function (RBF) is introduced to replace dot product operation in a high-dimensional space, so that the calculated amount can be greatly reduced, and the RBF is easy to realize the optimization process of the SVM because the center of each basis function of the RBF corresponds to the support vector one by one, and the support vector and the weight can be obtained through an algorithm. Thus, the time series least squares support vector machine (LS-SVM) predictor output is:
the output of a predictor of a time series least square support vector machine (LS-SVM) is a standard displacement prediction value, each intermediate node corresponds to a support vector, and the input is x1,x2,…xnFor the etalon shift time series history data, alphaiIs the network weight.
(5) Time series DRNN neural network predictor design
The time series DRNN neural network predictor is a dynamic regression neural network with feedback and the ability of adapting to time-varying characteristics, the network can more directly and vividly reflect the dynamic variation performance of the displacement of the standard device, can more accurately predict the displacement of the standard device and can more accurately predict the DRNN neural network predictor of the time seriesThe network predictor is a 3-layer network structure of 12-25-1, and the hidden layer is a regression layer. In the time series DRNN neural network predictor, I is set as [ I ]1(t),I2(t),…,In(t)]The displacement continuous values of the standard device at 12 different moments in a period of time are time series DRNN neural network input vectors, wherein Ii(t) is the input of the ith neuron of the input layer of the standard device displacement prediction model DRNN neural network at the t moment, and the output of the jth neuron of the regression layer is Xj(t)，Sj(t) is the sum of the j-th regression neuron inputs, f (-) is a function of S, then O (t) is the output of the DRNN neural network as the normalizer displacement predictor. The output of the DRNN network prediction model is:
(6) wavelet neural network fusion device design
The input of the wavelet neural network fusion device is the output values of a time sequence DRNN neural network predictor and a time sequence least square support vector machine (LS-SVM) predictor, the wavelet neural network fusion device realizes high-precision fusion of the output values of the time sequence DRNN neural network predictor and the time sequence least square support vector machine (LS-SVM) predictor, the displacement fusion accuracy of the standard device is improved, the output value of the wavelet neural network fusion device serves as a feedback prediction value of an intelligent standard device displacement control system, and prediction control of the intelligent standard device displacement controller is realized. The wavelet Neural network fusion device is a standard device displacement prediction fusion model constructed based on WNN (wavelet Neural networks) theoretical basis, and the wavelet Neural network is a feedforward network provided by taking a wavelet function as an excitation function of a neuron and combining an artificial Neural network. The expansion, translation factors and connection weights of the wavelets in the wavelet neural network fusion prediction model are adaptively adjusted in the optimization process of the error energy function. The input of the wavelet neural network fusion device is the output signals of a time series least square support vector machine (LS-SVM) predictor and a time series DRNN neural network predictor which can be expressed as a one-dimensional vector xi(i-1, 2, …, n) and the output signal is a etalon displacement prediction fusion value expressed as yk(k ═ 1,2, …, m), the calculation formula of the fusion value of the wavelet neural network fusion device output layer is:
in the formula omegaijInputting the connection weight between the i node of the layer and the j node of the hidden layer,
as wavelet basis functions, bjIs a shift factor of the wavelet basis function, ajScale factor, omega, of wavelet basis functionsjkThe connection weight between the node of the hidden layer j and the node of the output layer k. The correction algorithm of the weight and the threshold of the wavelet neural network fusion device in the patent adopts a gradient correction method to update the network weight and the wavelet basis function parameters, so that the output of the wavelet neural network fusion device is continuously close to the expected output of the displacement of the standard device.
The above embodiments are merely illustrative of the technical concepts and features of the present invention, and the purpose of the embodiments is to enable those skilled in the art to understand the contents of the present invention and to implement the present invention, and not to limit the scope of the present invention. All equivalent changes and modifications made according to the spirit of the present invention should be covered within the protection scope of the present invention.
Claims (6)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010694962.7A CN112013203B (en) | 2020-07-18 | 2020-07-18 | A Pipe Network Detection System Based on DRNN Neural Network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010694962.7A CN112013203B (en) | 2020-07-18 | 2020-07-18 | A Pipe Network Detection System Based on DRNN Neural Network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112013203A CN112013203A (en) | 2020-12-01 |
| CN112013203B true CN112013203B (en) | 2021-09-24 |
Family
ID=73499264
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010694962.7A Active CN112013203B (en) | 2020-07-18 | 2020-07-18 | A Pipe Network Detection System Based on DRNN Neural Network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112013203B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114839881B (en) * | 2022-06-17 | 2023-06-30 | 淮阴工学院 | Intelligent garbage cleaning and environmental parameter big data Internet of things system |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE69323943D1 (en) * | 1992-09-10 | 1999-04-22 | Deere & Co | Controller for a machine, especially for a combine harvester with a neural network |
| CN1273516A (en) * | 1997-08-28 | 2000-11-15 | 特性计算公司 | Method and system for automated detection of clustered microcalcifications from digital mammograms |
| CN102445597A (en) * | 2010-10-14 | 2012-05-09 | 刘一峰 | Harmonic measurement device |
| CN103376742A (en) * | 2012-04-24 | 2013-10-30 | 中国科学院合肥物质科学研究院 | CPG control system of wall-climbing robot imitating flexible structure of feet of gecko |
| CN103486134A (en) * | 2013-09-27 | 2014-01-01 | 江苏大学 | Construction method for decoupling controller of alternating-current hybrid magnetic bearing |
| CN107085732A (en) * | 2017-05-12 | 2017-08-22 | 淮阴工学院 | Intelligent monitoring system for ammonia in cowshed environment based on wireless sensor network |
| CN108091111A (en) * | 2017-11-20 | 2018-05-29 | 淮阴工学院 | A kind of oil truck oil and gas leakage intelligent early-warning system |
| CN110186454A (en) * | 2019-06-12 | 2019-08-30 | 海南大学 | A kind of filtering method of underwater inspection device and its attitude transducer |
| CN209856616U (en) * | 2019-01-25 | 2019-12-27 | 河南理工大学 | A walking mechanism of a supporting self-locking pipeline detection robot |
| WO2020212644A1 (en) * | 2019-04-18 | 2020-10-22 | Tampere University Foundation Sr | A home automation system |
Family Cites Families (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101387128B (en) * | 2004-12-14 | 2010-11-03 | 株式会社东芝 | Rainwater drain support system and method |
| CN101886743B (en) * | 2010-06-30 | 2012-10-17 | 东北大学 | A method and device for positioning a submarine pipeline robot |
| CN103615663B (en) * | 2013-11-11 | 2015-10-07 | 江苏师范大学 | Inner condition of pipeline measuring robots |
| CN104191429B (en) * | 2014-07-28 | 2016-04-27 | 南京航空航天大学 | The mixing control method of a kind of tendon driving device hand position and tendon tension force and control device |
| CN105204343B (en) * | 2015-10-13 | 2018-05-15 | 淮阴工学院 | The Nano electro-mechanical system backstepping control methods inputted with output constraint and dead band |
| US10443873B1 (en) * | 2016-02-03 | 2019-10-15 | Alarm.Com Incorporated | Energy reduction |
| CN105856238B (en) * | 2016-06-12 | 2018-11-16 | 江苏若博机器人科技有限公司 | Four core, six axis is wirelessly transferred the quick natural gas line robot control system of crawler type |
| JP6685404B2 (en) * | 2016-08-18 | 2020-04-22 | 株式会社日立産機システム | Piping network leak detection system, leak detection device used therefor, and leak detection method |
| CN106683222A (en) * | 2016-12-12 | 2017-05-17 | 中国人民解放军装甲兵技术学院 | Detection device and detection method for technical performances of engineering vehicle |
| WO2018112833A1 (en) * | 2016-12-22 | 2018-06-28 | Intel Corporation | Efficient transferring of human experiences to robots and other autonomous machines |
-
2020
- 2020-07-18 CN CN202010694962.7A patent/CN112013203B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE69323943D1 (en) * | 1992-09-10 | 1999-04-22 | Deere & Co | Controller for a machine, especially for a combine harvester with a neural network |
| CN1273516A (en) * | 1997-08-28 | 2000-11-15 | 特性计算公司 | Method and system for automated detection of clustered microcalcifications from digital mammograms |
| CN102445597A (en) * | 2010-10-14 | 2012-05-09 | 刘一峰 | Harmonic measurement device |
| CN103376742A (en) * | 2012-04-24 | 2013-10-30 | 中国科学院合肥物质科学研究院 | CPG control system of wall-climbing robot imitating flexible structure of feet of gecko |
| CN103486134A (en) * | 2013-09-27 | 2014-01-01 | 江苏大学 | Construction method for decoupling controller of alternating-current hybrid magnetic bearing |
| CN107085732A (en) * | 2017-05-12 | 2017-08-22 | 淮阴工学院 | Intelligent monitoring system for ammonia in cowshed environment based on wireless sensor network |
| CN108091111A (en) * | 2017-11-20 | 2018-05-29 | 淮阴工学院 | A kind of oil truck oil and gas leakage intelligent early-warning system |
| CN209856616U (en) * | 2019-01-25 | 2019-12-27 | 河南理工大学 | A walking mechanism of a supporting self-locking pipeline detection robot |
| WO2020212644A1 (en) * | 2019-04-18 | 2020-10-22 | Tampere University Foundation Sr | A home automation system |
| CN110186454A (en) * | 2019-06-12 | 2019-08-30 | 海南大学 | A kind of filtering method of underwater inspection device and its attitude transducer |
Non-Patent Citations (2)
| Title |
|---|
| 基于深度神经网络的车辆检测系统;宋佳兴等;《传动技术》;20191231;第33卷(第04期);第3-9页 * |
| 基于轻型卷积神经网络的火焰检测方法;严云洋等;《山东大学学报(工学版)》;20200318;第50卷(第2期);第100-107页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112013203A (en) | 2020-12-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112000008B (en) | Pipeline monitoring system based on neural network | |
| Lennox et al. | Industrial application of neural networks—an investigation | |
| Khalid et al. | A neural network controller for a temperature control system | |
| Akpan et al. | Nonlinear model identification and adaptive model predictive control using neural networks | |
| US5704011A (en) | Method and apparatus for providing multivariable nonlinear control | |
| CN102902203A (en) | Time series prediction and intelligent control combined online parameter adjustment method and system | |
| CN112013203B (en) | A Pipe Network Detection System Based on DRNN Neural Network | |
| CN108555914B (en) | A DNN neural network adaptive control method based on tendon-driven dexterous hand | |
| CN117472099A (en) | Filling machine liquid level control method based on fuzzy PID composite regulation | |
| CN114355848B (en) | Tension detection and intelligent control system | |
| Uçak | A Runge–Kutta neural network-based control method for nonlinear MIMO systems | |
| Leephakpreeda | Flow-sensorless control valve: neural computing approach | |
| Bamimore et al. | Offset-free neural network-based nonlinear model predictive controller design using parameter adaptation | |
| Lahtinen | Identification of fuzzy controller for use with a falling-film evaporator | |
| CN110643485A (en) | An automatic control system and method for temperature field in grain fermentation process with intelligent prediction compensator | |
| Mao et al. | Simulation of liquid level cascade control system based on genetic Fuzzy PID | |
| CN102662324A (en) | Non-linear model predication control method of tank reactor based on on-line support vector machine | |
| Wang et al. | Structure and weight optimization of neural network based on CPA-MLR and its application in naphtha dry point soft sensor | |
| RU83728U1 (en) | INTELLIGENT MANAGEMENT SYSTEM FOR ROBOT-MANIPULATOR MOVEMENT | |
| Sowmya et al. | A fuzzy control scheme for nonlinear process | |
| Xiong et al. | Fuzzy fractional-order PID control of the winding web tension in flexible electronic roll-to-roll manufacturing process | |
| Huang et al. | Neural network prediction modeling for a continuous, snack food frying process | |
| Jaleel et al. | Identification of ethane-ethylene distillation column using neural network and ANFIS | |
| Kästner et al. | Towards adaptive and efficient bottling plants in a cyber physical production system environment | |
| Lu et al. | Predictive control based on recurrent neural network and application to plastic injection molding processes |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20201201 Assignee: SUZHOU YOUPAKE PACKAGING MATERIALS Co.,Ltd. Assignor: HUAIYIN INSTITUTE OF TECHNOLOGY Contract record no.: X2021980013449 Denomination of invention: A pipe network detection system based on DRNN neural network Granted publication date: 20210924 License type: Common License Record date: 20211130 |
|
| EE01 | Entry into force of recordation of patent licensing contract | ||
| EC01 | Cancellation of recordation of patent licensing contract |
Assignee: SUZHOU YOUPAKE PACKAGING MATERIALS Co.,Ltd. Assignor: HUAIYIN INSTITUTE OF TECHNOLOGY Contract record no.: X2021980013449 Date of cancellation: 20220804 |
|
| EC01 | Cancellation of recordation of patent licensing contract | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20240407 Address after: 510000 A203, wisdom building, e-Park Creative Park, Guangzhou Yuzhu wisdom Valley, 32 kengtian street, Maogang village, Huangpu District, Guangzhou City, Guangdong Province Patentee after: Guangzhou Yonghao Technology Co.,Ltd. Country or region after: China Address before: 223005 Jiangsu Huaian economic and Technological Development Zone, 1 East Road. Patentee before: HUAIYIN INSTITUTE OF TECHNOLOGY Country or region before: China |
|
| TR01 | Transfer of patent right |