CN113325714A - Intelligent water-saving spraying system and method based on matrix completion and neural network PID - Google Patents
Intelligent water-saving spraying system and method based on matrix completion and neural network PID Download PDFInfo
- Publication number
- CN113325714A CN113325714A CN202110644492.8A CN202110644492A CN113325714A CN 113325714 A CN113325714 A CN 113325714A CN 202110644492 A CN202110644492 A CN 202110644492A CN 113325714 A CN113325714 A CN 113325714A
- Authority
- CN
- China
- Prior art keywords
- matrix
- water
- data
- completion
- neural network
- 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.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
- G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
- G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
- G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Feedback Control In General (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses an intelligent water-saving spraying system and method based on matrix completion and neural network PID (proportion integration differentiation). the spraying system comprises a water economizer and a master control terminal, wherein a plurality of water economizers are uniformly installed in the same land area, the water economizers are all in communication connection with the master control terminal, and the master control terminal realizes the control of the water economizers; the water economizer comprises a humidity acquisition module, a single chip microcomputer, a wireless communication module, a steering engine and a water control blade; the water control blade is arranged in the water spray pipe and is connected with an output shaft of the steering engine; the single chip microcomputer is respectively connected with the humidity acquisition module and the steering engine and is in communication connection with the master control terminal through the wireless communication module. The spraying system and the spraying method provided by the embodiment of the invention can effectively solve the problem of water resource waste and achieve the aim of saving water.
Description
Technical Field
The invention belongs to the technical field of water-saving control, and particularly relates to an intelligent water-saving spraying system and method based on matrix completion and neural network PID.
Background
Due to the action of atmospheric sunlight and wind heat, water is evaporated in soil all the time, the humidity value of the soil is possibly reduced rapidly, flowers and crops cannot obtain sufficient water for growth, and sometimes, plant roots and stems are soaked due to too much watering or water resources are wasted. Moreover, in the case of a large-area flower house, there may be a case where a plurality of plants are grown together and crops are cultivated in a mixed manner (corn-soybean), each plant has a different desired humidity value and growth condition, so that the difference of watering amount must be controlled to maintain a water resource-plant growth balance. The traditional watering device can only adopt a unified watering mode of 'flooding irrigation-irrigation', which can cause water shortage of plants, can cause a great deal of waste of water resources in most cases, and can not achieve a plurality of problems such as local conditions and the like for different plants.
Disclosure of Invention
The invention aims to provide an intelligent water-saving spraying system and method based on matrix completion and neural network PID (proportion integration differentiation), and aims to solve the problem that in the prior art, a traditional watering device adopts a watering mode of 'flooding-irrigation' to cause a large amount of waste of water resources.
In order to achieve the purpose, the invention adopts the following technical scheme:
an intelligent water-saving spraying system based on matrix completion and neural network PID comprises a water economizer and a master control terminal, wherein a plurality of water economizers are uniformly installed in the same land area, the plurality of water economizers are in communication connection with the master control terminal, and the master control terminal realizes control over the plurality of water economizers;
the water economizer comprises a humidity acquisition module, a single chip microcomputer, a wireless communication module, a steering engine and a water control blade; the water control blade is arranged in the water spray pipe and is connected with an output shaft of the steering engine;
the single chip microcomputer is respectively connected with the humidity acquisition module and the steering engine and is in communication connection with the master control terminal through the wireless communication module.
The embodiment of the invention provides another technical scheme that:
an intelligent water-saving spraying method based on matrix completion and neural network PID comprises the following steps:
acquiring humidity data of a current land area;
constructing an original matrix according to the humidity data, and visualizing the constructed original matrix;
performing data completion on the visualized original matrix to obtain an optimized final completion matrix;
inputting the optimized final completion matrix result value into a BP neural network for fuzzification processing to obtain a fuzzified final completion matrix value;
and inputting the fuzzy final completion matrix numerical value into a PID control algorithm fused with the CMAC cerebellar neural network for controlling the relevant parameters of the water economizer.
Furthermore, after the humidity data of the current land area are obtained, the correctness of the humidity data is firstly checked, and the data correctness rate is guaranteed to be more than 98%.
Further, the method for checking correctness is as follows:
detecting whether the humidity data has missing items or not and whether NaN, infinity and infinity data exist or not; if missing entries exist and data of NaN, infinity and infinity exist, the general control terminal sends an instruction to a corresponding water economizer to resend the data;
and after the corresponding water economizer resends the data, checking again, and considering that the read data are all correct when the data accuracy is more than 98% along with the increase of the checking times.
Further, the original matrix construction method comprises the following steps:
the length and the width of the current land area are used as rows and columns of a matrix, a coordinate system is established for one vertex of the matrix, the position of the water saver is reflected into the coordinate matrix, each coordinate point is composed of an abscissa x, an ordinate y and specific data z, and the z reflects soil humidity information.
Further, the method for performing data completion on the visualized original matrix comprises the following steps:
modeling matrix data in the visualization matrix as a tensor;
performing tensor matrix completion on missing frequency spectrum data in the tensor to obtain a pre-completion matrix after data is primarily completed;
and inputting the data of the pre-completion matrix into a BP neural network to learn to obtain new reference data, and finally obtaining an optimized final completion matrix.
Further, the BP neural network is constructed in the following manner: and (3) taking the read historical humidity data as an input variable, taking the humidity data after historical adjustment as an output variable, and establishing an activation function by determining the number of the neurons and the hidden layers so as to establish the BP neural network.
Further, drawing an original thermodynamic diagram according to the optimized final completion matrix, and displaying the original thermodynamic diagram on a software interface.
Further, the PID control algorithm fused with the CMAC cerebellar neural network comprises the following steps:
u(k)＝un(k)+up(k)
in the formula, alphaiVector selection for binary, c CMAC network generalization parameter, un(k) Generating a corresponding output, u, for the CMACp(k) The output generated by the conventional controller PID, and u (k) is the total control output;
at the end of each control cycle, the corresponding CMAC output u is calculatedn(k) And compared with the total control output u (k), the weight is corrected, and the learning process is carried out, so that the difference between the total control output and the output of the CMAC is minimum;
after CMAC learning, the total control output is generated by the CMAC.
Further, the adjustment objective function of the CMAC is:
w(k)＝w(k-1)+Δw(k)+α(w(k)-w(k-1))
in the formula, eta is the network learning rate, eta belongs to (0, 1), alpha is the inertia quantity, alpha belongs to (0, 1), and k is the learning times;
when is tiedWhen the system starts to operate, setting w to be 0; at this time un＝0；u＝upThe system is controlled by a conventional controller; through the learning of CMAC network, the output control quantity u of PID is madep(k) Gradually zero, output control u of CMAC generationn(k) Gradually approaches the controller total output u (k).
The invention has the following beneficial effects:
the spraying system and the spraying method provided by the embodiment of the invention can effectively solve the problem of water resource waste, acquire real-time water volume data of land through the humidity sensor connected with the water pipe, and transmit the data to the computer by virtue of the singlechip. The computer processes data through a program which is designed to be fused with a neural network and a PID algorithm, and then transmits the data to the steering engine to adjust the deflection angle of the blades in the water pipe, so that the water spraying amount is controlled according to the real water demand of different blocks, the plant needs are met, waste is avoided, and the purpose of saving water is achieved.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this application, illustrate embodiments of the invention and, together with the description, serve to explain the invention and not to limit the invention. In the drawings:
fig. 1 is a schematic structural diagram of a water economizer in an embodiment of the invention.
Fig. 2 is a schematic view showing the installation of the water saver in the embodiment of the present invention.
FIG. 3 is a schematic diagram of a spraying method in an embodiment of the present invention.
Fig. 4 is a schematic diagram of singular value changes of a matrix after tensor completion according to an embodiment of the present invention.
FIG. 5 is a comparison diagram of a matrix simulation after tensor completion in an embodiment of the present invention. Wherein, (a) is an actual humidity spectrum situation diagram, (b) is a schematic diagram of random missing 20% humidity data, and (c) is a humidity spectrum situation diagram filled by adopting ALM.
Fig. 6 is a schematic diagram of designed operating software, in which the upper right corner is an original humidity data acquisition distribution diagram, the left side is a supplemented real-time garden humidity data diagram, and the lower right corner also has correlation coefficients that can be manually set, such as target level, supplementation rate, acquisition time interval, data compression rate, and the like.
Wherein: 1, a humidity acquisition module; 2, a singlechip; 3, a battery; 4, a steering engine; 5 water control blades.
Detailed Description
The present invention will be described in detail below with reference to the embodiments with reference to the attached drawings. It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict.
The following detailed description is exemplary in nature and is intended to provide further details of the invention. Unless otherwise defined, all technical terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of exemplary embodiments according to the invention.
Firstly, as shown in fig. 1 and 2, an aspect of the embodiments of the present invention provides an intelligent water-saving spraying system based on matrix completion and neural network PID, which includes a water-saving device and a master control terminal, wherein a plurality of water-saving devices are uniformly installed in the same land area, and are all in communication connection with the master control terminal; the master control terminal realizes the control of a plurality of water-saving devices and realizes the water-saving function.
Specifically, in the embodiment, the water economizer comprises a humidity acquisition module, a single chip microcomputer, a wireless communication module, a steering engine and a water control blade; the water control blade is arranged in the water spray pipe and is connected with an output shaft of the steering engine; the single chip microcomputer is respectively connected with the humidity acquisition module and the steering engine and is in communication connection with the master control terminal through the wireless communication module.
In this embodiment, the single chip microcomputer is selected according to actual requirements by using an STM32 minimum system board F103C8T6, or a 51 single chip microcomputer minimum system board STC89C51, AT89C51 of Atmel company or other programmable microprocessors. The wireless communication module is an HC-12 wireless communication module 433M and is responsible for data communication between the single chip microcomputer and the master control terminal. The humidity acquisition module is a humidity sensor. The steering engine model is SG 90. The singlechip RXD is connected with the TXD of the HC-12 wireless communication module 433M, the singlechip TXD is connected with the RXD of the HC-12 wireless communication module 433M, a 23A/12V alkaline battery is used for supplying power to the singlechip mainboard, and the grounding end is conducted through the shell of the water economizer.
The water control blade is connected with an actual common water pipe, after the receiving end of the wireless communication module reads angle information, the steering engine can control the rotation of the connecting gear by a specific angle, and the water control blade rotates immediately to control the flow velocity of the water pipe. It should be noted that the plug-in water economizer in this embodiment is suitable for all types of common water spraying pipes, and any type available in the market may be adopted, and is not described herein again. Before working, a common water spraying pipe is punched out of a matching port, a water control blade and a fixed interface of a water economizer are installed on the matching port, three detection needles of a humidity sensor are inserted into the position 7cm below soil, so that the humidity information of the soil is detected, and then the humidity information is converted into an electric signal to be transmitted to a single chip microcomputer.
In another aspect of the embodiments of the present invention, an intelligent water-saving spraying method based on matrix completion and neural network PID is provided, which specifically includes:
and step S1, selecting rectangular land areas, and determining the number of water-saving devices according to budget. And determining the inserting density and the inserting position of the water economizer according to the existing water spraying pipe distribution structure of the current land area. When the water economizer is installed, the water economizers are inserted at equal intervals.
Preferably, in one embodiment of the present invention, not all of the spray pipes are installed with water savers from the viewpoint of further cost saving. For some users who can not bear the cost of most of the water saving devices, the water saving devices can be selectively inserted in a certain range to obtain the maximum economic benefit.
Specifically, when the water economizer is installed, an inserting opening is punched at a proper position of a water spraying pipe, the water control blade is arranged in the water spraying pipe and is fixed with an output shaft of the steering engine through the inserting opening, and the inserting opening is sealed by using a rubber ring and glass cement; inserting three detection needles of a humidity sensor into a position 7-8 cm below soil, and fixing the probes; and detecting whether the working performance of the probe is good or not, and determining that the probe of the humidity sensor can detect the humidity information of the soil.
Step S2, when water saving adjustment is started, the humidity sensor transmits the collected humidity electric signal to the single chip microcomputer through the RC485 module; the single chip microcomputer transmits the newly read soil humidity data to the master control terminal through the HC-12 wireless communication module.
Specifically, in this embodiment, one master control terminal performs PID control on a plurality of water savers, and a multi-transmission and one-reception working mode is adopted.
Preferably, in an embodiment of the present invention, after the single chip reads the humidity data, the read humidity data is stored in the E2PROM module, so as to prevent data loss due to power failure.
And step S3, after the master control terminal reads the humidity data sent by the wireless communication module, firstly, the data correctness is checked, and the data correctness is ensured to be more than 98%.
The method comprises the following specific steps:
in step S31, it is detected whether or not there are missing entries in the humidity data and whether or not there are NaN, + ∞ and- ∞ data. If missing entries exist and data of NaN, + ∞andinfinity exist, the general control terminal sends an instruction to a corresponding water saver to resend the data.
Specifically, in this embodiment, the wireless communication module of the water saver adopts a full duplex operating mode of multiple transmission and one reception, and the transmitted data is encoded in an 8-bit binary form. And an odd check/even check mode is adopted to ensure the data transmission rate, and the data transmission rate is respectively received and transmitted through TXD and RXD of the wireless communication module.
And step S32, after the corresponding water saver resends the data, checking again, and considering that the read data are all correct when the data correctness rate is more than 98% along with the increase of the checking times.
And step S4, constructing an original matrix according to the humidity data with the accuracy rate of more than 98% obtained in the step S3, and visualizing the constructed original matrix.
Specifically, the way of constructing the original matrix is as follows: the length and width of the selected rectangular land area are used as rows and columns of the matrix, a coordinate system is established for one vertex of the matrix, the position of the water saver is reflected into the coordinate matrix, each coordinate point (x, y, z) is composed of an abscissa x, an ordinate y and specific data z, and in the embodiment, z reflects soil humidity information.
Specifically, the original matrix can be visualized as shown in fig. 4, wherein the visualized matrix is just mapped to the physical position of the selected land area, and the color depth represents the size of the specific data, so as to obtain the missing matrix information and the visualized matrix.
Step S5, after the original matrix constructed in step S4 is visualized, as shown in fig. 4, the humidity data collected by the water saver is only displayed in a dotted distribution on the graph, and data information at a position between the water saver and the water saver is difficult to collect in many places, so in this embodiment, prediction and completion processing is performed on data that cannot be collected.
Specifically, in this embodiment, the prediction and completion processing method for the data that cannot be collected is as follows:
step S51, modeling the matrix data in the visualization matrix as a tensor.
S52, performing tensor matrix completion such as ALM and SVT on missing frequency spectrum data in the tensor, and the specific principle is that incomplete matrixes are recovered by utilizing the low rank and sparsity of the matrixes, missing data are predicted, and therefore all soil information is read; and then a pre-completion matrix after the data is primarily completed is obtained.
And step S6, inputting the data of the pre-completion matrix obtained in the step S5 into a BP neural network to learn to obtain new reference data, and finally obtaining an optimized final completion matrix.
Specifically, in this embodiment, the BP neural network is constructed in the following manner: and (3) taking the read historical humidity data as an input variable, taking the humidity data after historical adjustment as an output variable, and establishing an activation function by determining the number of the neurons and the hidden layers so as to establish the BP neural network.
Preferably, in an embodiment of the present invention, there is no historical humidity data when the BP neural network is first utilized for optimization, so that no adjustment of the BP neural network is required, but on the other hand, historical data can be manually input, and the estimation and input of the artificial matrix can be performed according to the historical experience of the user, the current season, and different requirements of each crop.
Preferably, in an embodiment of the present invention, an original thermodynamic diagram is drawn according to the optimized final completion matrix, and the original thermodynamic diagram is displayed on a software interface, so as to reflect the real-time humidity level.
And step S7, inputting the optimized final completion matrix result value into a BP neural network for fuzzification processing to obtain a fuzzified final completion matrix value, and summarizing conditions possibly occurring in the process of adjusting the angle of the water control blade by the steering engine and corresponding control strategies into a fuzzy control table.
And S8, inputting a PID control algorithm fused with the CMAC (cerebellar neural network) by using the fuzzified final completion matrix value obtained in the step S7, and controlling related parameters of the water economizer.
Specifically, the control algorithm is designed as follows:
1) determining CMAC structure, determining node numbers M and Q of input layer and hidden layer, and giving weighting coefficient of each layer
η and the relation coefficient α, k are selected.
2) By sampling, r (k) and y (k), e (k) r (k) -y (k) are calculated. r (k) is the original input quantity, e (k) is the PID controller output quantity after feedback, and y (k) is the output quantity.
3) And e (k) is fuzzified, and the obtained result is used as the input of the BP neural network. The output of the output layer is the adjustable parameter of the steering engine PID controller.
Specifically, the feedforward feedback control is realized by the CMAC feedforward control and the composite control of the CMAC and the PID. The control algorithm is as follows:
u(k)＝un(k)+up(k)
in the formula, alphaiThe vector is selected for the binary system and,c is a CMAC network generalization parameter, un(k) Generating a corresponding output, u, for the CMACp(k) The output generated by the conventional controller PID, u (k) is the overall control output.
At the end of each control cycle, the corresponding CMAC output u is calculatedn(k) And compared with the total control output u (k), the weight is corrected, and the learning process is carried out, so that the difference between the total control output and the output of the CMAC is minimized.
After CMAC learning, the total control output of the system is generated by the CMAC. The adjustment objective function of the CMAC is:
w(k)＝w(k-1)+Δw(k)+α(w(k)-w(k-1))
in the formula, η is the network learning rate, η belongs to (0, 1), α is the inertia, α belongs to (0, 1), and k is the learning frequency.
When the system starts to operate, setting w to be 0; at this time un＝0；u＝upThe system is controlled by a conventional controller. Through the learning of CMAC network, the output control quantity u of PID is madep(k) Gradually zero, output control u of CMAC generationn(k) Gradually approaches the controller total output u (k).
And step S9, calculating a deviation value with the actual/ideal value according to the adjustable parameters of the steering engine PID controller, judging whether the deviation value meets a set standard or an ideal standard in the mind of a user, if the deviation value does not meet the standard, setting k to k +1, and repeating the step S8.
And S10, the master control terminal returns the target humidity value matrix of each water economizer and the angle of the steering engine blade to be adjusted to the single chip microcomputer through the transparent transmission function, the single chip microcomputer transmits the steering engine angle information to the SG90 steering engine, the steering engine is linked with the water control blade to rotate to the target angle, and the water control blade blocks the water flow of the water spray pipe.
Specifically, in this embodiment, a one-transmission and multiple-reception operating mode is adopted to transmit the data to each water economizer, and the water economizer receives the data through the wireless communication module and returns the data to the single chip microcomputer. Thereby quantitatively regulating and controlling the soil humidity to a specific value.
Preferably, in an embodiment of the present invention, after the first adjustment, the process of watering-interval time-reading again may be performed, and the process of S1, S2, S3. is repeated at a time interval set by a person, so as to gradually adjust to a desired level, thereby achieving the purpose of saving water, energy and money.
Thirdly, the validity of the method of the embodiment of the invention is verified by combining a specific simulation experiment:
in the embodiment of the invention, after experimental data are drawn up, Matlab program simulation is carried out, partial simulation results are shown, and the effectiveness and superiority of the method can be seen as follows:
in the present invention, the completion of missing data is based on the low rank nature of the matrix. In order to verify the correlation of data used in an experiment, firstly, the low rank property of a matrix is verified, the known data is the matrix, singular values of the matrix are arranged from large to small, the change is shown in fig. 4, most of the singular values tend to 0, and the matrix has the characteristic of low rank:
moreover, in order to verify the superiority of the ALM algorithm, the filling effect thereof is compared through simulation:
when the humidity sensors of different devices acquire the real-time humidity of the land and are transmitted by the single chip microcomputer, different conditions of each block can be displayed in the page. Because the sampling range of the sampling device is smaller than the spraying range of the spray head of the device, the related data in the range is supplemented to 50% -100% in different degrees according to the requirement. According to the characteristics of plants, the degree of water spraying amount can be artificially selected according to the weather and sunshine conditions of the day. And through the processing of a PID algorithm, the system can calculate the water demand of different blocks according to the data, transmit the water demand to the steering engine, and adjust the deflection angle of the blades in the water pipe, so that the water spraying amount is controlled according to the real water demand of different blocks. The time interval of data sampling, data compression rate, etc. can all be adjusted on the page.
In order to verify the effect of the application, the growth conditions of main plants and the water demand of the whole growth period in a national main crop water demand map are searched.
By using the principle of 'control test', two flat open lands are supposed to exist, and the conditions such as soil, temperature and the like are appropriate. In order to pursue the universality of the results, nine crops, three fruits, three common agricultural products and three flowers are designed and planted. One group is provided with a common spray irrigation system (the water demand is the searched data), the other group is provided with the water-saving spray pipe system, a completion coefficient and a water-saving coefficient are set according to the plant type and the growth condition, and the deflection angle of the blade is regulated and controlled.
Under the condition of mass purchase, a single spraying system only needs about 10 yuan, the spraying radius is about 5 meters, more than 60 square meters can be controlled, only 100 yuan is needed for 1 mu of land, and the water-saving efficiency is greatly improved.
It will be appreciated by those skilled in the art that the invention may be embodied in other specific forms without departing from the spirit or essential characteristics thereof. The embodiments disclosed above are therefore to be considered in all respects as illustrative and not restrictive. All changes which come within the scope of or equivalence to the invention are intended to be embraced therein.
Claims (10)
1. The intelligent water-saving spraying system based on matrix completion and neural network PID is characterized by comprising a water economizer and a master control terminal, wherein a plurality of water economizers are uniformly installed in the same land area, the plurality of water economizers are all in communication connection with the master control terminal, and the master control terminal realizes control over the plurality of water economizers;
the water economizer comprises a humidity acquisition module, a single chip microcomputer, a wireless communication module, a steering engine and a water control blade; the water control blade is arranged in the water spray pipe and is connected with an output shaft of the steering engine;
the single chip microcomputer is respectively connected with the humidity acquisition module and the steering engine and is in communication connection with the master control terminal through the wireless communication module.
2. An intelligent water-saving spraying method based on matrix completion and neural network PID is characterized by comprising the following steps:
acquiring humidity data of a current land area;
constructing an original matrix according to the humidity data, and visualizing the constructed original matrix;
performing data completion on the visualized original matrix to obtain an optimized final completion matrix;
inputting the optimized final completion matrix result value into a BP neural network for fuzzification processing to obtain a fuzzified final completion matrix value;
and inputting the fuzzy final completion matrix numerical value into a PID control algorithm fused with the CMAC cerebellar neural network for controlling the relevant parameters of the water economizer.
3. The intelligent water-saving spraying method based on matrix completion and neural network PID according to claim 2, characterized in that after the humidity data of the current land area is obtained, the correctness of the humidity data is checked to ensure that the data correctness is more than 98%.
4. The intelligent water-saving spraying method based on matrix completion and neural network PID according to claim 3, characterized in that the correctness checking method is:
detecting whether the humidity data has missing items or not and whether NaN, infinity and infinity data exist or not; if missing entries exist and data of NaN, infinity and infinity exist, the general control terminal sends an instruction to a corresponding water economizer to resend the data;
and after the corresponding water economizer resends the data, checking again, and considering that the read data are all correct when the data accuracy is more than 98% along with the increase of the checking times.
5. The intelligent water-saving spraying method based on matrix completion and neural network PID according to claim 2, characterized in that the original matrix construction method is:
the length and the width of the current land area are used as rows and columns of a matrix, a coordinate system is established for one vertex of the matrix, the position of the water saver is reflected into the coordinate matrix, each coordinate point is composed of an abscissa x, an ordinate y and specific data z, and the z reflects soil humidity information.
6. The intelligent water-saving spraying method based on matrix completion and neural network PID according to claim 2, characterized in that the method for performing data completion on the visualized original matrix is as follows:
modeling matrix data in the visualization matrix as a tensor;
performing tensor matrix completion on missing frequency spectrum data in the tensor to obtain a pre-completion matrix after data is primarily completed;
and inputting the data of the pre-completion matrix into a BP neural network to learn to obtain new reference data, and finally obtaining an optimized final completion matrix.
7. The intelligent water-saving spraying method based on matrix completion and neural network PID according to claim 6, characterized in that BP neural network construction mode is: and (3) taking the read historical humidity data as an input variable, taking the humidity data after historical adjustment as an output variable, and establishing an activation function by determining the number of the neurons and the hidden layers so as to establish the BP neural network.
8. The intelligent water-saving spraying method based on matrix completion and neural network PID as claimed in claim 2, characterized in that according to the optimized final completion matrix, an original thermodynamic diagram is drawn and displayed on a software interface.
9. The intelligent water-saving spraying method based on matrix completion and neural network PID according to claim 2, characterized in that the PID control algorithm fused with the CMAC cerebellar neural network is:
u(k)＝un(k)+up(k)
in the formula, alphaiVector selection for binary, c CMAC network generalization parameter, un(k) Generating a corresponding output, u, for the CMACp(k) The output generated by the conventional controller PID, and u (k) is the total control output;
at the end of each control cycle, the corresponding CMAC output u is calculatedn(k) And compared with the total control output u (k), the weight is corrected, and the learning process is carried out, so that the difference between the total control output and the output of the CMAC is minimum;
after CMAC learning, the total control output is generated by the CMAC.
10. The intelligent water-saving spraying method based on matrix completion and neural network PID according to claim 9, characterized in that the adjustment objective function of CMAC is:
in the formula, eta is the network learning rate, eta belongs to (0, 1), alpha is the inertia quantity, alpha belongs to (0, 1), and k is the learning times;
when the system starts to operate, setting w to be 0; at this time un＝0；u＝upThe system is controlled by a conventional controller; through the learning of CMAC network, the output control quantity u of PID is madep(k) Gradually zero, output control u of CMAC generationn(k) Gradually approaches the controller total output u (k).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110644492.8A CN113325714B (en) | 2021-06-09 | 2021-06-09 | Intelligent water-saving spraying system and method based on matrix completion and neural network PID |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110644492.8A CN113325714B (en) | 2021-06-09 | 2021-06-09 | Intelligent water-saving spraying system and method based on matrix completion and neural network PID |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113325714A true CN113325714A (en) | 2021-08-31 |
| CN113325714B CN113325714B (en) | 2023-02-24 |
Family
ID=77420341
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110644492.8A Active CN113325714B (en) | 2021-06-09 | 2021-06-09 | Intelligent water-saving spraying system and method based on matrix completion and neural network PID |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113325714B (en) |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1951170A (en) * | 2005-10-17 | 2007-04-25 | 王光强 | Remote controlled automated irrigation system based on public communication network and control method thereof |
| CN101167436A (en) * | 2006-10-24 | 2008-04-30 | 中兴通讯股份有限公司 | Intelligent irrigation system and method based on wireless detector |
| CN102323834A (en) * | 2011-05-27 | 2012-01-18 | 西安天茂节能科技有限公司 | A kind of intelligent water-saving control system |
| JP2014153949A (en) * | 2013-02-08 | 2014-08-25 | Hiroshima Univ | Non-linear pid controller using gmdh and gmdh weighting factor determination method |
| CN105830874A (en) * | 2016-04-30 | 2016-08-10 | 邵阳学院 | Accurate water-saving irrigation remote wireless intelligent control system |
| US20170061452A1 (en) * | 2015-08-31 | 2017-03-02 | Wal-Mart Stores, Inc. | System for forecasting using low-rank matrix completion and method therefor |
| CN107197439A (en) * | 2017-06-01 | 2017-09-22 | 南京邮电大学 | Wireless sensor network locating method based on matrix completion |
| CN109856973A (en) * | 2019-02-25 | 2019-06-07 | 山东省农业机械科学研究院 | A kind of Technique for Controlling Greenhouse Environment and system based on fuzzy neural network |
| CN110463587A (en) * | 2019-09-21 | 2019-11-19 | 河北工程大学 | Scale greenhouse water-saving irrigation system based on LoRa and GA-BP |
| CN110572789A (en) * | 2019-08-12 | 2019-12-13 | 东北大学秦皇岛分校 | Wireless sensor network high-dimensional data completion method based on Hankel transformation |
| CN111401553A (en) * | 2020-03-12 | 2020-07-10 | 南京航空航天大学 | Missing data filling method and system based on neural network |
| CN112003327A (en) * | 2020-08-31 | 2020-11-27 | 南方电网能源发展研究院有限责任公司 | Neural network adaptive control-based grid-connected pre-synchronization control method and system |
-
2021
- 2021-06-09 CN CN202110644492.8A patent/CN113325714B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1951170A (en) * | 2005-10-17 | 2007-04-25 | 王光强 | Remote controlled automated irrigation system based on public communication network and control method thereof |
| CN101167436A (en) * | 2006-10-24 | 2008-04-30 | 中兴通讯股份有限公司 | Intelligent irrigation system and method based on wireless detector |
| CN102323834A (en) * | 2011-05-27 | 2012-01-18 | 西安天茂节能科技有限公司 | A kind of intelligent water-saving control system |
| JP2014153949A (en) * | 2013-02-08 | 2014-08-25 | Hiroshima Univ | Non-linear pid controller using gmdh and gmdh weighting factor determination method |
| US20170061452A1 (en) * | 2015-08-31 | 2017-03-02 | Wal-Mart Stores, Inc. | System for forecasting using low-rank matrix completion and method therefor |
| CN105830874A (en) * | 2016-04-30 | 2016-08-10 | 邵阳学院 | Accurate water-saving irrigation remote wireless intelligent control system |
| CN107197439A (en) * | 2017-06-01 | 2017-09-22 | 南京邮电大学 | Wireless sensor network locating method based on matrix completion |
| CN109856973A (en) * | 2019-02-25 | 2019-06-07 | 山东省农业机械科学研究院 | A kind of Technique for Controlling Greenhouse Environment and system based on fuzzy neural network |
| CN110572789A (en) * | 2019-08-12 | 2019-12-13 | 东北大学秦皇岛分校 | Wireless sensor network high-dimensional data completion method based on Hankel transformation |
| CN110463587A (en) * | 2019-09-21 | 2019-11-19 | 河北工程大学 | Scale greenhouse water-saving irrigation system based on LoRa and GA-BP |
| CN111401553A (en) * | 2020-03-12 | 2020-07-10 | 南京航空航天大学 | Missing data filling method and system based on neural network |
| CN112003327A (en) * | 2020-08-31 | 2020-11-27 | 南方电网能源发展研究院有限责任公司 | Neural network adaptive control-based grid-connected pre-synchronization control method and system |
Non-Patent Citations (1)
| Title |
|---|
| 史新华 等: "《建筑水暖及空调工程施工质量控制》", 31 May 2006, 中国计划出版社 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113325714B (en) | 2023-02-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104460582B (en) | A kind of Internet of Things intelligent irrigation fertilising control method and system based on fuzzy control | |
| Adenugba et al. | Smart irrigation system for environmental sustainability in Africa: An Internet of Everything (IoE) approach | |
| CN108094370A (en) | Control method of irrigation and device | |
| CN107945042A (en) | A kind of plant growth irrigation decision control system | |
| CN106651012A (en) | Crop transpiration prediction method based on improved extreme learning machine | |
| CN107330804A (en) | A kind of wisdom water conservancy management and control cloud platform and method | |
| CN110036881A (en) | A kind of wisdom agricultural irrigation systems and its control method | |
| CN107173184A (en) | A kind of Agricultural Intelligent System irrigation system and method | |
| CN110909933B (en) | Agricultural drought rapid diagnosis and evaluation method coupling crop model and machine learning language | |
| CN103823371A (en) | Neural network model-based agricultural precise fertilization system and fertilization method thereof | |
| Suciu et al. | Efficient IoT system for precision agriculture | |
| CN114365614A (en) | Water and fertilizer accurate regulation and control method, intelligent equipment and system based on Internet of things | |
| CN115661547B (en) | Knowledge-graph-based plant irrigation maintenance method and system | |
| CN110163254A (en) | A kind of cucumber green house yield intelligent Forecasting device based on recurrent neural network | |
| CN115530054A (en) | Irrigation control method and device, electronic equipment and storage medium | |
| CN110579961B (en) | Three-dimensional planting-oriented garden intelligent water supply method and system | |
| CN110754333A (en) | Irrigation scheduling method suitable for irrigation area | |
| CN113554522A (en) | Vineyard accurate drip irrigation control system based on dynamic neural network | |
| CN111459033A (en) | Grey prediction fuzzy PID control method and equipment for water and fertilizer irrigation | |
| CN115633622A (en) | Intelligent orchard irrigation system and method | |
| CN115879613A (en) | Agricultural industry park load optimization scheduling method considering carbon transaction | |
| CN113325714B (en) | Intelligent water-saving spraying system and method based on matrix completion and neural network PID | |
| CN109116827A (en) | Heliogreenhouse water-fertilizer integral control method of irrigation and device based on Internet of Things | |
| CN113837891A (en) | Balanced and efficient water resource allocation method for large-area agricultural irrigation area coping with climate change | |
| CN114219673A (en) | Agricultural cloud service system based on Internet of things |
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 |