CN113008440B - Flexible liquid injection sensor detection method based on genetic algorithm optimization neural network - Google Patents
Flexible liquid injection sensor detection method based on genetic algorithm optimization neural network Download PDFInfo
- Publication number
- CN113008440B CN113008440B CN202110262434.9A CN202110262434A CN113008440B CN 113008440 B CN113008440 B CN 113008440B CN 202110262434 A CN202110262434 A CN 202110262434A CN 113008440 B CN113008440 B CN 113008440B
- Authority
- CN
- China
- Prior art keywords
- liquid injection
- neural network
- flexible
- output
- layer
- 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
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01L—MEASURING FORCE, STRESS, TORQUE, WORK, MECHANICAL POWER, MECHANICAL EFFICIENCY, OR FLUID PRESSURE
- G01L11/00—Measuring steady or quasi-steady pressure of a fluid or a fluent solid material by means not provided for in group G01L7/00 or G01L9/00
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/086—Learning methods using evolutionary algorithms, e.g. genetic algorithms or genetic programming
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Physiology (AREA)
- Measuring Fluid Pressure (AREA)
Abstract
The invention discloses a flexible liquid injection sensor detection method based on a genetic algorithm optimized neural network, which relates to the technical field of intelligent flexible pressure sensing equipment.
Description
Technical Field
The invention relates to the technical field of intelligent flexible pressure sensing equipment, in particular to a flexible liquid injection sensor detection method based on a genetic algorithm optimized neural network.
Background
The dynamic change of the stress of the surrounding rock is an important reason for causing various geological disasters. In recent years, with the development of economy in China and the continuous increase of coal mining depth and strength, the surrounding rock stress distribution change rule is more and more complex, and the probability of geological disasters caused by the change is gradually increased. At present, the monitoring equipment for surrounding rock stress change is mainly rigid equipment such as a borehole stress meter, the change of surrounding rock stress cannot be sensitively sensed, and the surrounding rock stress change in a fixed direction can only be monitored, so that the surrounding rock rule is not accurately mastered, and hidden dangers are left for geological disasters. Therefore, the relation between the stress change of the surrounding rock, the deformation of the flexible sensor and the output of the pressure gauge is found, and the rule of the change of the supporting pressure is found, so that the method has important significance for the safety production of the coal mine.
Disclosure of Invention
In order to solve the technical problems, the invention provides the flexible liquid injection sensor detection method based on the genetic algorithm optimization neural network, which can improve the detection efficiency, the automation degree and the intelligence of the flexible liquid injection sensor in a reply manner and has stronger practicability.
In order to achieve the purpose, the technical scheme adopted by the invention is as follows:
a flexible liquid injection sensor detection method based on a genetic algorithm optimization neural network comprises the steps of setting a plurality of flexible liquid injection sensors, building and training a neural network prediction model, setting a plurality of drill holes on surrounding rocks by a genetic algorithm, installing the flexible liquid injection sensors in each drill hole, detecting stress at different positions on the surrounding rocks, dividing each flexible liquid injection sensor into 4 sections, applying different pressures to the sections respectively, performing simulation acquisition on the stress of the surrounding rocks, taking the stress value of the surrounding rocks measured at each section 1 as the input value of the neural network prediction model, taking the reading of a pressure gauge as the output value of the neural network prediction model to obtain the high-precision approximate nonlinear function relationship between the stress change of the surrounding rocks and the deformation of the pressure gauge, performing parameter optimization on the neural network prediction model through the genetic algorithm to realize accurate and rapid prediction of output data, and adopting a high-performance chip as a control center controller by the control system, and connecting the flexible liquid injection sensors with a data bus of a field bus network in a communication mode to realize the communication between the output pressure gauge and a control center.
The neural network prediction model construction and training method is realized by the following steps:
the first stage is as follows: the method comprises the steps of building a neural network prediction model and collecting data, wherein the neural network prediction model is composed of an input layer, a hidden layer and an output layer, neurons of the input layer are set to have four sections of diameters d1, d2, d3 and d4 corresponding to 4 sections of equal distribution of a flexible liquid injection sensor, neurons of the output layer are set to be pressure readings p output by a pressure gauge, the number of the neurons of the hidden layer is generally determined by an empirical formula, different surrounding rock stresses are simulated and applied through experimental equipment, required input and output data are collected, n groups of data are recorded, and corresponding sample data are obtained;
and a second stage: and training the neural network prediction model, wherein the training process mainly comprises two processes of forward propagation of signals and backward propagation of errors, when the error between actual output and target output does not meet the preset precision requirement, the neural network can continuously adjust the weight, update the network until the error is smaller than the preset precision, and the training is finished.
The number of layers of the hidden layer is set to be 1, and the number of neurons of the hidden layer is set to be 9.
The genetic algorithm is designed and realized by the following steps:
s1, selecting binary codes;
s2, initializing a group;
s3, selecting an error performance index function to define a fitness function, evaluating an adaptive value and storing an optimal chromosome;
s4, roulette selection operation, single-point crossing operation and variation operation with variation rate of 0.000-0.1 are adopted;
s5, generating a new population, and recalculating the adaptive value of each chromosome;
and S6, judging whether the termination condition is met, when the algorithm reaches the termination condition, terminating the algorithm, outputting a result, and if the termination condition is not met, returning to S3.
The flexible liquid injection sensor comprises at least one flexible pressure sensing device, a liquid injection device, a conveying device and a pressure measuring device, wherein the liquid injection device is connected with the flexible pressure sensing device through the conveying device and used for injecting emulsion into the flexible pressure sensing device, and the pressure measuring device is communicated with the conveying device and the flexible pressure sensing device respectively and used for displaying the pressure of the emulsion through a pressure gauge.
The flexible pressure sensing device comprises a pipe body, a sealing layer, a rubber layer and a joint, wherein the sealing layer and the rubber layer are arranged outside the pipe body respectively, one end of the pipe body is sealed, the other end of the pipe body is connected with the joint, a liquid injection hole is formed in the middle of the joint, an exhaust hole is formed in the side face of the joint, and the liquid injection hole and the exhaust hole are communicated with an inner cavity of the pipe body.
The injection device comprises a pump, a flowmeter and an injection gun, the conveying device is a rubber tube, and the pump, the flowmeter, the injection gun and the rubber tube are connected in sequence and used for conveying emulsion.
The pressure measuring device comprises a pressure gauge and a three-way valve, one end of the three-way valve is communicated with the pressure gauge, and the other two ends of the three-way valve are respectively communicated with the flexible pressure sensing device and the liquid injection device.
STM32, DSP, CPLD or FPGA high-performance chips are used as controllers.
The invention designs a flexible liquid injection sensor which is arranged in a plurality of drill holes on surrounding rock, detects stress values of different positions on the surrounding rock and acquires data, utilizes the nonlinear mapping capability of an artificial neural network, trains a neural network prediction model to enable the network to approach a nonlinear function with high precision, obtains the relation between stress change of the surrounding rock and deformation of a pressure gauge according to input stress value data and pressure gauge output value data, realizes prediction of the output data, introduces a genetic algorithm to carry out parameter optimization on a BP neural network prediction model, improves the precision of the network and the training speed, greatly improves the precision and rapidity of the detected data, solves the problem that the neural network is easy to fall into local optimum to a great extent, greatly improves the detection efficiency of the system and the automation degree of a surrounding rock stress detection system through a field bus communication mode of a plurality of master stations, has strong practicability, and is suitable for large-scale popularization and application.
Drawings
FIG. 1 is a schematic diagram of the principle of construction and training of a neural network prediction model of the present invention;
FIG. 2 is a schematic diagram of a flexible fluid injection sensor according to the present invention;
fig. 3 is a schematic structural view of the flexible pressure-sensing device of the present invention.
Detailed Description
The invention is described in detail below with reference to the following figures and specific embodiments:
the flexible liquid injection sensor detection method based on the genetic algorithm optimized neural network comprises the steps of setting a plurality of flexible liquid injection sensors, building and training a neural network prediction model, setting a plurality of drill holes on a surrounding rock by the genetic algorithm, installing the flexible liquid injection sensors in each drill hole, detecting stress at different positions on the surrounding rock, dividing each flexible liquid injection sensor into 4 sections, applying different pressures respectively, carrying out simulated acquisition on the stress of the surrounding rock, taking the stress value of the surrounding rock measured at each section as the input value of the neural network prediction model, taking the reading of a pressure gauge as the output value of the neural network prediction model, obtaining the high-precision approximate nonlinear function relation between the stress change of the surrounding rock and the deformation of the pressure gauge, carrying out parameter optimization on the neural network prediction model by the genetic algorithm, and realizing accurate and rapid prediction on-output data, wherein the liquid injection sensor detection system adopts a high-performance chip as a control center controller, and connects the flexible sensors with a data bus of a field bus network in a communication mode, so as to realize the communication between the output pressure gauge and a control center.
As shown in fig. 1, in this embodiment, as a preferred mode, the method for constructing and training the neural network prediction model is implemented by the following steps:
the first stage is as follows: the method comprises the steps of building a neural network prediction model and collecting data, wherein the neural network prediction model is composed of an input layer, a hidden layer and an output layer. The neuron of the input layer is set to four sections of diameters d1, d2, d3 and d4 corresponding to 4 sections of the flexible liquid injection sensor, and the neuron of the output layer is set to a pressure reading p output by a pressure gauge, so that the number of input nodes is 4, and the number of output nodes is 1. The number of neurons in the hidden layer is generally determined by an empirical formula. Different surrounding rock stresses are applied through simulation of experimental equipment, required input and output data are collected, n groups of data are recorded, and corresponding sample data are obtained.
And a second stage: training of a neural network prediction model mainly comprises two processes of forward propagation of signals and backward propagation of errors. And during forward propagation, inputting the input sample data acquired in the first stage into an input layer of the neural network, and transmitting the input sample data into an output layer after the hidden layer processing. If the actual output of the output layer has a larger error value with the target output, the error is transferred to a reverse propagation stage. The error back propagation is to reversely transmit the output error to the input layer by layer through the hidden layer and distribute the error to all units of each layer, thereby obtaining the error signal of each layer of units, and the error signal is used as the basis for correcting the weight of each unit. And when the error between the actual output and the target output does not meet the preset precision requirement, the neural network can continuously adjust the weight and update the network until the error is smaller than the preset precision, and the training is finished.
Specifically, in this embodiment, the number of layers of the hidden layer is set to 1, and the number of neurons in the hidden layer is set to 9.
Preferably, the genetic algorithm described in this embodiment is designed and implemented by the following steps:
s1, selecting binary codes;
s2, initializing a group;
s3, selecting an error performance index function to define a fitness function, evaluating an adaptive value and storing an optimal chromosome; the fitness function is a measure for representing the quality of an individual object in a problem by considering the adaptation degree of a biological individual to the environment, and is generally determined according to an objective function of an actual problem;
s4, roulette selection operation, single-point crossing operation and variation operation with variation rate of 0.000-0.1 are adopted;
s5, generating a new population, and recalculating the adaptive value of each chromosome;
and S6, judging whether a termination condition is met, when the algorithm reaches the termination condition, terminating the algorithm, and outputting a result. The termination conditions are generally of three types: an optimal or near optimal solution has been obtained; the fitness is saturated; continued iteration does not result in better individuals. The set maximum evolution algebra is reached, and the artificially set evolution algebra is generally 100-500 generations. If the termination condition is not satisfied, the process returns to S3.
As shown in fig. 2-3, in a preferred embodiment, the flexible liquid injection sensor includes at least one flexible pressure sensing device 1, a liquid injection device 2, a conveying device 3, and a pressure measuring device 4, where the liquid injection device 1 is connected to the flexible pressure sensing device 1 through the conveying device 2 for injecting emulsion into the flexible pressure sensing device 1, and the pressure measuring device 4 is respectively communicated with the conveying device 2 and the flexible pressure sensing device 1, and is used for displaying pressure of the emulsion therein through a pressure gauge. When the stress of the surrounding rock is detected, a plurality of drill holes are generally formed in the surrounding rock, and a flexible pressure sensing device 1 is installed in each drill hole to detect the stress at different positions on the surrounding rock. Flexible pressure sensing device 1 is the cartridge type structure for in the drilling of country rock is packed into, after the emulsion that has predetermined pressure is injected into, flexible pressure sensing device 1 can with the laminating of drilling inner wall, and the stress that any direction was received to the drilling inner wall all can be transmitted to flexible pressure sensing device 1's surface, detects the pressure of emulsion in the flexible pressure sensing device 1 through pressure measuring device 4, just can detect the stress in the drilling of different positions on the country rock.
Specifically, in this embodiment, the flexible pressure sensing device 1 includes a tube 11, a sealing layer 12, a rubber layer 13, and a joint 14, the sealing layer 12 and the rubber layer 13 are respectively disposed outside the tube 11, one end of the tube 11 is closed, the other end of the tube is connected to the joint 14, a liquid injection hole 141 is disposed in the middle of the joint 14, an exhaust hole 142 is disposed on the side of the joint, and both the liquid injection hole 141 and the exhaust hole 142 are communicated with an inner cavity of the tube 11. The rubber layer 13 is formed by pressing and weaving high polymer material rubber and steel wires.
Specifically, in this embodiment, the liquid injection device 2 includes a pump 21, a flow meter 22 and a liquid injection gun 23, the conveying device 3 is provided as a rubber tube, and the pump 21, the flow meter 22, the liquid injection gun 23 and the rubber tube are connected in sequence for conveying the emulsion.
Specifically, in this embodiment, the pressure measuring device 4 includes a pressure gauge and a three-way valve 41, one end of the three-way valve 41 is communicated with the pressure gauge, and the other two ends are respectively communicated with the flexible pressure sensing device 3 and the liquid injection device 2.
As a preferable mode, in this embodiment, an STM32, a DSP, a CPLD, or an FPGA high-performance chip is used as the controller, which can rapidly and continuously acquire system parameters, perform high-precision data processing and arithmetic operations, and timely and accurately provide a control signal.
The field bus communication mode can be wired or wireless. Wired communication modes include optical fiber, shielded twisted pair, coaxial cable, RS485, RS232, profibus, profinet, FF, modbus, CAN bus and the like, and wireless communication modes include WiFi, zigbee, loRa, infrared and the like.
The operation of the detection method of the present invention is briefly described as follows:
s1, averagely dividing a flexible liquid injection sensor into 4 sections according to the length;
s2, applying different pressures to each section respectively to simulate the change of the stress of the surrounding rock and observe the output condition of the pressure dial plate;
s3, respectively recording the pressure value of 4 sections during each pressure application and a corresponding pressure gauge display value to obtain sample data of the neural network;
s4, dividing sample data into a training set and a test set, training the neural network model by using the training set, and testing the trained model by using the test set;
and S5, combining the trained neural network model with genetic algorithm optimization parameters for predicting the output value of the pressure gauge.
The specific detection method comprises the following steps:
1) The neural network prediction model is structurally designed and comprises an input layer, a hidden layer and an output layer, wherein neurons of the input layer are 4 sections of diameters of the flexible liquid injection sensor, neurons of the output layer are pressure readings output by a pressure gauge, nodes of the input layer are 4, and nodes of the output layer are 1. The number of hidden layers is generally selected to be a single hidden layer. It has been theoretically demonstrated that a three-layer feedforward neural network can approximate any continuous function with any accuracy. Thus implying a number of layers of 1. In the network design process, the determination of the number of the neurons of the hidden layer is very important. The number of hidden layer neurons is too large, so that the network calculation amount is increased, and the overfitting problem is easy to generate; if the number of the neurons is too small, the network performance is affected and the expected effect cannot be achieved. The number of hidden layer neurons in the network has a direct link to the complexity of the real problem, the number of neurons in the input and output layers, and the setting of the expected error. At present, for the hiddenThe determination of the number of neurons in the stratum has no clear formula, but only a few empirical formulas, and the final determination of the number of neurons still needs to be determined according to experience and multiple experiments. The invention refers to the following empirical formula on the problem of selecting the number of hidden layer neurons:
in the formula: j is the number of hidden layer neurons, i and k are the number of input layer and output layer neurons, respectively, and a is generally a constant between [1,10 ]. In the invention, the number of the neurons in the hidden layer is taken as 9.
2) The training process of the BP neural network prediction model mainly comprises two processes of forward propagation of signals and backward propagation of errors. And during forward propagation, inputting the input sample data acquired in the first stage into an input layer of the neural network, and transmitting the input sample data into an output layer after the hidden layer processing. If the actual output of the output layer has a larger error value with the expected output, the error is transmitted to the back propagation stage. The error back propagation is to reversely transmit the output error to the input layer by layer through the hidden layer and distribute the error to all units of each layer, thereby obtaining the error signal of each layer of units, and the error signal is used as the basis for correcting the weight of each unit. And when the error between the actual output and the target output does not meet the preset precision requirement, the neural network can continuously adjust the weight and update the network until the error is smaller than the preset precision, and the training is finished.
And collecting sample data, measuring the pressure of 4 sections of the flexible liquid injection sensor, and recording the pressure value of 4 sections and the corresponding display value of the pressure gauge when the pressure is applied each time to obtain the sample data of the neural network. The sample is divided into a training set and a test set. Input sample data is recorded as: d (m) = [ d = [) 1 (m),d 2 (m),K d n (m)]Wherein n =1,2, \8230, 200, m is the training and learning times.
Initializing data, initializing weight V between input layer and hidden layer ij And the weight W between the hidden layer and the output layer jk And a threshold value a for the hidden layer and a threshold value b for the output layer, given a learning rateAnd an activation function.
Hidden layer input:
hidden layer output calculation: from the input samples, the output of the hidden layer is computed forward. The hidden layer output is:
and (3) input of an output layer:
and (4) output layer output calculation, and further calculating the output of the output layer according to the output of the hidden layer. The output of the output layer is:
and the error signal is obtained by comparing the output of the output layer with the expected value.
The error signals are: e.g. of the type k ＝T k -P k 。
Error performance index:
where T is the desired output and P is the actual output.
Partial derivative of error function to output layer input:
the output layer inputs the partial derivative of the connection weight between the hidden layer and the output layer:
partial derivative of the error function to the connection weight between the hidden layer and the output layer:
partial derivative of error function to hidden layer input:
the hidden layer inputs the partial derivative of the connection weight between the input layer and the hidden layer:
partial derivative of the error function to the connection weight between the input layer and the hidden layer:
the network error is a function of the weights of the layers, so adjusting the weights can change the error E. The principle of adjusting the weight is to make the error decrease continuously.
Updating the weight between the hidden layer and the output layer:
W jk (m+1)＝W jk (m)+ΔW jk (m)＝W jk (m)+ηgδ 1 (m)gOo j (m)
updating the weight between the input layer and the hidden layer:
V ij (m+1)＝V ij (m)+ΔV ij (m)＝V ij (m)+ηgδ 2 (m)gd i (m)。
i is the input layer neuron number, j is the hidden layer neuron number, k is the output layer neuron number, and η is the learning rate, which is set to 0.01 in this example. V ij ，W jk Respectively, the connection weights between the input layer and the hidden layer and between the hidden layer and the output layer, and f is a continuously derivable Sigmoid function.
And (3) error judgment: and judging whether the error reaches the upper limit of the preset precision or the learning times. If the requirement is met, the neural network training is finished, otherwise, the learning is returned again.
Through the global optimization of the genetic algorithm, the problem that the BP neural network is easy to fall into local optimization can be solved to a great extent.
The test set data is used for verifying the neural network prediction model, and the result shows that the difference between the predicted value and the true value output by the neural network to the pressure gauge is small, and the relative error is small, so that the model can accurately reflect the nonlinear mapping relation between input and output, can accurately describe the internal relation between the surrounding rock stress and the pressure gauge output data, and can accurately predict the pressure gauge output data.
It is to be understood that the above description is not intended to limit the present invention, and the present invention is not limited to the above examples, and those skilled in the art may make various changes, modifications, additions and substitutions within the spirit and scope of the present invention.
Claims (7)
1. A flexible liquid injection sensor detection method based on a genetic algorithm optimization neural network is characterized by comprising the steps of setting a plurality of flexible liquid injection sensors, constructing and training a neural network prediction model, optimizing the neural network prediction model through a genetic algorithm, and controlling the neural network prediction model, wherein a plurality of drill holes are formed in a surrounding rock, each drill hole is provided with a flexible liquid injection sensor, and the stress at different positions on the surrounding rock is detected;
the flexible liquid injection sensor comprises at least one flexible pressure sensing device (1), a liquid injection device (2), a conveying device (3) and a pressure measuring device (4), wherein the liquid injection device (1) is connected with the flexible pressure sensing device (1) through the conveying device (2) and used for injecting emulsion into the flexible pressure sensing device (1), and the pressure measuring device (4) is respectively communicated with the conveying device (2) and the flexible pressure sensing device (1) and used for displaying the pressure of the emulsion in the flexible pressure sensing device through a pressure gauge;
the pressure measuring device (4) comprises a pressure gauge and a three-way valve (41), one end of the three-way valve (41) is communicated with the pressure gauge, and the other two ends of the three-way valve are respectively communicated with the flexible pressure sensing device (3) and the liquid injection device (2);
the method comprises the steps of equally dividing a flexible pressure sensing device of each flexible liquid injection sensor into 4 sections, applying different pressures respectively, carrying out analog acquisition on surrounding rock stress, taking a surrounding rock stress value measured at each 1 section as an input value of a neural network prediction model, taking a reading of a pressure gauge as an output value of the neural network prediction model to obtain a high-precision approximate nonlinear function relation between surrounding rock stress change and pressure gauge deformation, carrying out parameter optimization on the neural network prediction model through a genetic algorithm to realize accurate and rapid prediction of output data, and connecting a plurality of flexible liquid injection sensors with a data bus of a field bus network in a communication mode by using a high-performance chip as a control center controller of a control system to realize communication between the output pressure gauge and the control center.
2. The flexible liquid injection sensor detection method based on the genetic algorithm optimization neural network is characterized in that the construction and training method of the neural network prediction model is realized by the following steps:
the first stage is as follows: the method comprises the steps of building a neural network prediction model and collecting data, wherein the neural network prediction model is composed of an input layer, a hidden layer and an output layer, neurons of the input layer are set to have four sections of diameters d1, d2, d3 and d4 corresponding to 4 sections of flexible liquid injection sensors, neurons of the output layer are set to be pressure readings p output by a pressure gauge, the number of the neurons of the hidden layer is generally determined by adopting an empirical formula, different surrounding rock stresses are simulated and applied through experimental equipment, required input and output data are collected, n groups of data are recorded, and corresponding sample data are obtained;
and a second stage: and training a neural network prediction model, wherein the training process mainly comprises two processes of forward propagation of signals and backward propagation of errors, when the error between actual output and target output does not meet the preset precision requirement, the neural network can continuously adjust the weight, update the network until the error is smaller than the preset precision, and the training is finished.
3. The flexible liquid injection sensor detection method based on the genetic algorithm optimization neural network as claimed in claim 2, wherein the number of hidden layers is set to 1, and the number of hidden layer neurons is set to 9.
4. The flexible liquid injection sensor detection method based on the genetic algorithm optimization neural network is characterized in that the genetic algorithm is designed and realized through the following steps:
s1, selecting binary codes;
s2, initializing a group;
s3, selecting an error performance index function to define a fitness function, evaluating an adaptive value, and storing an optimal chromosome;
s4, selecting operation by roulette, single-point crossing operation and variation operation with variation rate of 0.000-0.1 are adopted;
s5, generating a new population, and recalculating the adaptive value of each chromosome;
and S6, judging whether the termination condition is met, when the algorithm reaches the termination condition, terminating the algorithm, outputting a result, and if the termination condition is not met, returning to S3.
5. The flexible liquid injection sensor detection method based on the genetic algorithm optimized neural network is characterized in that the flexible pressure sensing device (1) comprises a pipe body (11), a sealing layer (12), a rubber layer (13) and a joint (14), wherein the sealing layer (12) and the rubber layer (13) are respectively arranged outside the pipe body (11), one end of the pipe body (11) is closed, the other end of the pipe body is connected with the joint (14), a liquid injection hole is formed in the middle of the joint (14), an exhaust hole is formed in the side face of the joint (14), and the liquid injection hole and the exhaust hole are both communicated with an inner cavity of the pipe body (11).
6. The flexible liquid injection sensor detection method based on the genetic algorithm optimization neural network is characterized in that the liquid injection device (2) comprises a pump (21), a flow meter (22) and a liquid injection gun (23), the conveying device (3) is arranged to be a rubber tube, and the pump (21), the flow meter (22), the liquid injection gun (23) and the rubber tube are connected in sequence and used for conveying emulsion.
7. The flexible liquid injection sensor detection method based on the genetic algorithm optimization neural network is characterized in that an STM32, DSP, CPLD or FPGA high-performance chip is adopted as a controller.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110262434.9A CN113008440B (en) | 2021-03-10 | 2021-03-10 | Flexible liquid injection sensor detection method based on genetic algorithm optimization neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110262434.9A CN113008440B (en) | 2021-03-10 | 2021-03-10 | Flexible liquid injection sensor detection method based on genetic algorithm optimization neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113008440A CN113008440A (en) | 2021-06-22 |
| CN113008440B true CN113008440B (en) | 2023-01-10 |
Family
ID=76404632
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110262434.9A Active CN113008440B (en) | 2021-03-10 | 2021-03-10 | Flexible liquid injection sensor detection method based on genetic algorithm optimization neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113008440B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116842854B (en) * | 2023-09-01 | 2023-11-07 | 山东科技大学 | Intelligent prediction and reducing pressure relief method for coal body stress based on optimized neural network |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104132761A (en) * | 2014-08-04 | 2014-11-05 | 中国矿业大学 | Multipoint coal and rock mass stress real-time monitoring device and method |
| CN106989849A (en) * | 2017-05-15 | 2017-07-28 | 山东科技大学 | Single hole coal and rock directional stress is distributed and the integral monitoring device of deformation and monitoring method |
| CN109372581A (en) * | 2018-12-14 | 2019-02-22 | 山东科技大学 | A kind of roof strata horizontal compression monitoring device and application method |
| CN111365051A (en) * | 2020-03-06 | 2020-07-03 | 广西交通设计集团有限公司 | Method for estimating stress of carbonaceous rock tunnel anchor rod based on transfer function of feedback algorithm |
Family Cites Families (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103616111A (en) * | 2013-12-09 | 2014-03-05 | 中国科学院武汉岩土力学研究所 | Small-hole thick-wall sleeve core stress relieving method |
| CN104502002A (en) * | 2014-12-29 | 2015-04-08 | 山东华硕能源科技有限公司 | Hydraulic force transfer type borehole stressmeter and operation method thereof |
| CN110359905A (en) * | 2019-06-13 | 2019-10-22 | 山东大学 | A kind of device and method obtaining rock reaction force based on artificial neural network |
| CN110457758B (en) * | 2019-07-16 | 2023-03-24 | 江西理工大学 | Method, device and system for predicting unstable phase of rock mass and storage medium |
| CN111076848A (en) * | 2019-12-27 | 2020-04-28 | 天津大学 | Pressure measuring device and method |
| CN111335921B (en) * | 2020-02-28 | 2021-07-30 | 东南大学 | Automatic grouting device for mounting single-point displacement meter and sectional grouting mounting method |
| CN112100927A (en) * | 2020-09-24 | 2020-12-18 | 湖南工业大学 | Prediction method for slope deformation and soft soil foundation settlement based on GA-BP neural network |
-
2021
- 2021-03-10 CN CN202110262434.9A patent/CN113008440B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104132761A (en) * | 2014-08-04 | 2014-11-05 | 中国矿业大学 | Multipoint coal and rock mass stress real-time monitoring device and method |
| CN106989849A (en) * | 2017-05-15 | 2017-07-28 | 山东科技大学 | Single hole coal and rock directional stress is distributed and the integral monitoring device of deformation and monitoring method |
| CN109372581A (en) * | 2018-12-14 | 2019-02-22 | 山东科技大学 | A kind of roof strata horizontal compression monitoring device and application method |
| CN111365051A (en) * | 2020-03-06 | 2020-07-03 | 广西交通设计集团有限公司 | Method for estimating stress of carbonaceous rock tunnel anchor rod based on transfer function of feedback algorithm |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113008440A (en) | 2021-06-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110807557B (en) | BP neural network-based drilling rate prediction method and BP neural network-based particle swarm optimization method | |
| CN109063247B (en) | Landslide disaster forecasting method based on deep belief network | |
| CN111425768A (en) | Oil pipeline oil leakage point and oil leakage rate detection method based on underwater sensor network | |
| US20040156681A1 (en) | Fluid regulation | |
| CN101319925A (en) | Coal gas measuring method by utilization of BP neural network | |
| CN106198870B (en) | A kind of anchor rod body damage location identification method based on neural network | |
| CN101504333B (en) | Progressive health monitoring method for cable structure median cord system based on strain monitoring | |
| CN104881583A (en) | Multi-means, dynamic and whole-process landslide prewarning method | |
| CN113008440B (en) | Flexible liquid injection sensor detection method based on genetic algorithm optimization neural network | |
| CN113408190B (en) | Surrounding rock deformation prediction method for highway tunnel construction period based on Bayes-LSTM model | |
| CN112879025A (en) | Earth pressure balance shield earth pressure intelligent regulation and control method based on GRU deep learning | |
| CN106568647B (en) | A kind of Strength Forecast of Concrete method neural network based | |
| CN114358434A (en) | Drilling machine drilling speed prediction method based on LSTM recurrent neural network model | |
| CN115983097A (en) | Coal seam gas extraction characteristic parameter rapid inversion method based on borehole extraction data | |
| CN114357750A (en) | Goaf water filling state evaluation method | |
| CN116415479A (en) | Continuous oil pipe fatigue life early warning method based on digital twinning | |
| CN117189713A (en) | Hydraulic system fault diagnosis method based on digital twin driving | |
| CN101782473B (en) | Progressive type healthy monitoring method of rope system based on angle monitoring during support settlement | |
| CN114912364A (en) | Natural gas well flow prediction method, device, equipment and computer readable medium | |
| CN104899642B (en) | Large deformation flexible body dynamic stress compensation method based on hybrid production style | |
| CN117973194A (en) | Parallel system-based oil-gas pipe network digital twin modeling method and system | |
| CN111751878B (en) | Method and device for predicting transverse wave speed | |
| CN117236191A (en) | Method for predicting physical property parameters of reservoir based on deep learning technology | |
| CN117090831A (en) | Hydraulic system fault diagnosis framework with twinning application layer | |
| CN102937036A (en) | Method and device for monitoring gas on basis of BP (back propagation) neural network |
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 | ||
| CB03 | Change of inventor or designer information |
Inventor after: Wu Shitao Inventor after: Tang Jianquan Inventor after: Ye Wei Inventor after: Yang Jie Inventor after: Yu Ruirui Inventor before: Wu Shitao Inventor before: Tang Jianquan Inventor before: Ye Wei Inventor before: Yang Jie Inventor before: Yu Ruirui |
|
| CB03 | Change of inventor or designer information | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |