CN114839881B - Intelligent garbage cleaning and environmental parameter big data Internet of things system - Google Patents
Intelligent garbage cleaning and environmental parameter big data Internet of things system Download PDFInfo
- Publication number
- CN114839881B CN114839881B CN202210696166.6A CN202210696166A CN114839881B CN 114839881 B CN114839881 B CN 114839881B CN 202210696166 A CN202210696166 A CN 202210696166A CN 114839881 B CN114839881 B CN 114839881B
- Authority
- CN
- China
- Prior art keywords
- neural network
- network model
- parameter
- environmental
- output
- 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
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
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)
Abstract
The invention discloses an intelligent garbage cleaning and environmental parameter big data Internet of things system, which consists of an environmental parameter acquisition and control platform and an environmental parameter processing and garbage cleaning subsystem, wherein the environmental parameter acquisition and control platform is used for detecting, adjusting and monitoring the environmental parameter; the environmental parameter big data processing subsystem realizes the control of environmental parameter processing and garbage cleaning; the invention effectively solves the problems that the existing environmental garbage cleaning has no influence on environmental quality due to nonlinearity, large hysteresis, large and complex environmental area and the like which are changed according to environmental parameters, and has no prediction on the environmental parameters, accurate detection on the environmental parameters and adjustment of a garbage cleaning device, thereby greatly influencing the environmental parameter prediction and environmental management.
Description
Technical Field
The invention relates to the technical field of automatic equipment for detecting and processing environmental parameters and cleaning garbage, in particular to an intelligent garbage cleaning and environmental parameter big data Internet of things system.
Background
The ecological environment is a prerequisite for human survival and social production, and is the root for promoting economic civilization construction. Under the conditions of increasingly tense natural resources and increasingly worsened ecological environment, all human beings should stand up to protect nature, respect nature and conform to the environmental awareness of nature. The environmental awareness of residents is poor, the cognition on garbage classification is backward, and serious environmental pollution and serious damage to ecological environment are caused. The method is characterized in that important monitoring is carried out aiming at the environmental pollution problem to be solved, a large amount of wastes are important factors causing ecological environmental pollution, a large amount of human and livestock excrement wastes which can be recycled and utilized can be directly removed, the pollution of the excrement to the environment is reduced, villagers engaged in livestock breeding can conduct operation safely, the influence on the environment and surrounding residents is not needed, recycling and reutilization of garbage are realized, an environment automatic monitoring and garbage cleaning system is built, and the authenticity, objectivity and timeliness of environment detection data are ensured. The environment monitoring and garbage cleaning device is designed by using a network and intelligent control technology, and the intelligent garbage cleaning and environmental parameter big data Internet of things system is invented to improve the environment monitoring and environmental cleaning level.
Disclosure of Invention
The invention provides an intelligent garbage cleaning and environmental parameter big data Internet of things system, which effectively solves the problems that the existing environmental garbage cleaning has no influence on environmental quality due to nonlinearity, large hysteresis, large and complex environmental area and the like which are changed according to environmental parameters, and has no prediction on the environmental parameters, accurate detection on the environmental parameters and adjustment on a garbage cleaning device, so that the prediction on the environmental parameters and the environmental management are greatly influenced.
The invention is realized by the following technical scheme:
the intelligent garbage collection and environment parameter big data Internet of things system consists of an environment parameter collection and control platform and an environment parameter processing and garbage collection subsystem, wherein the environment parameter collection and control platform is used for detecting and managing environment parameters; the environment parameter processing and garbage cleaning subsystem realizes the processing of environment parameters and the adjustment of the garbage cleaning device, and improves the environment management efficiency and benefit.
The invention further adopts the technical improvement scheme that:
the environment parameter acquisition and control platform consists of a detection node, a control node, a gateway node, a field monitoring end, a cloud platform and a mobile end APP, wherein the detection node acquires environment parameters, the environment parameters are uploaded to the cloud platform through the gateway node, data provided by the cloud platform are utilized for the mobile end APP, the mobile end APP can monitor the environment parameters and adjust external equipment of the control node in real time through environment information provided by the cloud platform, the detection node and the control node are responsible for acquiring the environment parameter information and controlling environment adjusting equipment, and bidirectional communication among the detection node, the control node, the field monitoring end, the cloud platform and the mobile end APP is realized through the gateway node, so that the environment parameter acquisition and the environment equipment control are realized; the structure of the environment parameter acquisition and control platform is shown in fig. 1.
The invention further adopts the technical improvement scheme that:
the environment parameter processing and garbage cleaning subsystem consists of a parameter detection module, an ANFIS neural network model, a parameter self-adjusting factor fuzzy controller, a PID controller, an LSTM neural network model, an NARX neural network controller and a fuzzy wavelet neural network model, wherein the outputs of a plurality of groups of ammonia, hydrogen sulfide and carbon dioxide sensors are used as the input of the corresponding parameter detection module, the outputs of a plurality of groups of temperature, humidity and wind speed sensors are used as the input of the corresponding parameter detection module, the outputs of 2 parameter detection modules and the fuzzy wavelet neural network model are used as the corresponding input of the ANFIS neural network model, the attitude errors and the error change rates of the garbage cleaning device, which are output by the ANFIS neural network model and output by the fuzzy wavelet neural network model, are respectively used as the input of the parameter self-adjusting factor fuzzy controller and the PID controller, the output of the PID controller is used as the LSTM neural network model, the outputs of the LSTM neural network model and the parameter self-adjusting factor fuzzy controller are respectively used as the corresponding input of the NARX neural network controller, the output of the NARX neural network controller is used as the yaw angle, pitch angle and roll angle control quantity, the output by the MPU6050 attitude sensor, the time sequence of the output by the corresponding sensor and the yaw angle control quantity are respectively used as the corresponding parameter detection module. The structure of the environment parameter processing and garbage cleaning subsystem is shown in fig. 2.
The invention further adopts the technical improvement scheme that:
the parameter detection module consists of an NARX neural network model, an Adaline neural network model, a K-means cluster classifier, a CNN convolution-LSTM neural network model, an AANN self-association neural network model of a Vague set and a beat delay line TDL; the method comprises the steps that a plurality of groups of parameter sensors sense time sequence parameter values of a detected environment to be respectively used as inputs of a corresponding NARX neural network model and an Adaline neural network model, differences of the NARX neural network model and the Adaline neural network model are used as fluctuation values of detected parameter levels, a plurality of time sequence parameter fluctuation values and a plurality of Adaline neural network model outputs are respectively used as inputs of a corresponding K-means cluster classifier, a plurality of types of time sequence parameter fluctuation values and Adaline neural network model outputs which are output by 2K-means cluster classifiers are respectively used as inputs of a corresponding CNN convolution-LSTM neural network model, a plurality of CNN convolution-LSTM neural network model outputs to be used as corresponding inputs of an AANN self-association neural network model of a Vague set, three parameters of the AANN self-association neural network model of the Vague set are respectively x, t and 1-f, x is a real value of the detected parameter, t is a reliability degree, f is an uncertainty degree, 1-f-t is a value, x, 1-f-t is a value and 1-tdf is a value, and the delay of the AANN self-association neural network model of the Vague set is output as a value of the AAGUE, and the AANN self association neural network model of the vaue set is output as a value of the AAGUE, and the AANN of the data of the detection model is. The parameter detection module is shown in fig. 3.
Compared with the prior art, the invention has the following obvious advantages:
1. aiming at the uncertainty and the randomness of the problems of sensor precision error, interference, measurement abnormality and the like in the parameter measurement process, the invention converts the parameter value measured by the sensor into the numerical form representation of the detection parameter Vague set through the parameter detection module, effectively processes the ambiguity, the dynamic property and the uncertainty of the sensor measurement parameter and improves the objectivity and the credibility of the sensor detection parameter.
2. The invention relates to an NARX neural network controller, which is a dynamic recursive network for establishing the NARX neural network controller by introducing a delay module of an LSTM neural network model and an output characteristic parameter of a parameter self-adjusting factor fuzzy controller and NARX neural network controller output feedback realization.
3. According to the invention, K-means is adopted to perform cluster analysis on a plurality of time series parameter fluctuation values and a plurality of Adaline neural network model output data, a cluster center obtained through the cluster analysis classifies the plurality of time series parameter fluctuation values and the plurality of Adaline neural network model output data to be respectively input as corresponding CNN convolution-LSTM neural network models, and the plurality of time series parameter fluctuation values and the plurality of Adaline neural network model output data of different types are respectively predicted by adopting the corresponding CNN convolution-LSTM neural network models, so that the accuracy of detecting and predicting the input data is improved.
4. In the CNN convolution-LSTM neural network model adopted by the invention, the CNN convolution neural network is a deep feed-forward neural network, the typical structure of the CNN convolution neural network is composed of an input layer, a convolution layer, a pooling layer and a full-connection layer, the CNN convolution neural network is used for performing operations such as convolution and pooling on input data, and local characteristics of the data are extracted by establishing a plurality of filters, so that robust characteristics with translational rotation invariance are obtained. The LSTM neural network comprises an input layer, a hidden layer and an output layer, wherein the memory units are added in each neural unit of the hidden layer, so that information on a time sequence can be controlled to be forgotten or output, the problems of gradient explosion and gradient disappearance in RNN are solved, the LSTM neural network is far better than RNN in processing long-sequence data, characteristic information on the output time sequence of the CNN convolutional neural network can be effectively extracted by the LSTM neural network, a CNN convolutional-LSTM neural network model can fully mine spatial characteristic relations among variables of the output data of the CNN convolutional neural network, and time sequence characteristic information of the output historical data of the CNN convolutional neural network is extracted, so that the CNN convolutional-LSTM neural network model has stronger learning and generalization capability.
5. The AANN self-association neural network model of the Vague set extracts the most representative low-dimensional subspace of the input parameter change system structure in the high-dimensional parameter space reflecting the output values of the multiple input CNN convolution-LSTM neural network models, noise and errors in the input CNN convolution-LSTM neural network model data are filtered effectively, decompression of the input CNN convolution-LSTM neural network model data is achieved through a bottleneck layer, a demapping layer and an output layer, three parameters output by the AANN self-association neural network model of the Vague set are x, t and 1-f respectively, x is the real value of a detected parameter, t is the credibility, f is the uncertainty, x, t and 1-f constitute the numerical value of the detected parameter Vague set is [ x, (t, 1-f) ], and accuracy and robustness of predicting the detected parameter are improved.
6. The ANFIS neural network model adopted by the invention is a fuzzy reasoning system based on a Takagi-Sugeno model, is a novel fuzzy reasoning system structure which organically combines fuzzy logic and a neural network, adopts a mixed algorithm of a back propagation algorithm and a least square method to adjust precondition parameters and conclusion parameters, and automatically generates If-Then rules. The ANFIS neural network model is taken as a very characteristic neural network, has the function of approximating any linear and nonlinear functions with any precision, and has the advantages of high convergence rate, small sample requirement, high operation speed, reliable result and good effect.
Drawings
FIG. 1 is an environmental parameter acquisition and control platform of the present invention;
FIG. 2 is an environmental parameter processing and garbage disposal subsystem of the present invention;
FIG. 3 is a diagram illustrating a parameter detection module according to the present invention;
FIG. 4 is a schematic diagram of a detection node according to the present invention;
FIG. 5 is a control node of the present invention;
fig. 6 is a gateway node of the present invention;
fig. 7 is a view of the field monitoring software of the present invention.
Detailed Description
The technical scheme of the invention is further described with reference to the accompanying drawings 1-7:
1. system overall functional design
The intelligent garbage cleaning and environmental parameter big data Internet of things system realizes the detection of environmental parameters and the control of environmental regulation equipment, and the system consists of an environmental parameter acquisition and control platform and an environmental parameter processing and garbage cleaning subsystem. The environment parameter acquisition and control platform comprises detection nodes, control nodes, gateway nodes, a field monitoring end, a cloud platform and a mobile end APP of environment parameters, wherein the detection nodes and the control nodes are constructed into a LoRa monitoring network in a self-organizing mode to realize LoRa communication among the detection nodes, the control nodes and the gateway nodes; the detection node sends the detected environmental parameters to the field monitoring end and the cloud platform through the gateway node, and the gateway node and the cloud platform provide bidirectional transmission channels of the environmental parameters and related control information between the field monitoring end and the mobile end APP. The mobile terminal APP is designed by adopting an open source framework APP provided by the smart cloud, and a cloud platform can be connected and remote detection and regulation functions based on the mobile terminal APP can be realized only by integrating an APP SDK provided by the smart cloud in the mobile terminal APP. The structure of the environment parameter acquisition and control platform is shown in figure 1.
2. Design of detection node
A large number of detection nodes 1 based on the LoRa sensor network are used as environment parameters and garbage cleaning device gesture sensing terminals, and the detection nodes realize information interaction with gateway nodes through the self-organizing LoRa network. The detection node comprises a sensor for collecting parameters of ambient humidity, temperature, wind speed, ammonia gas, hydrogen sulfide and carbon dioxide, a corresponding signal conditioning circuit, an MPU6050 posture sensor, an STM32 microprocessor and a LoRa communication module SX1278; the software of the detection node mainly realizes the collection and pretreatment of LoRa communication and environmental parameters. The software adopts the C language programming, the compatibility degree is high, the working efficiency of software design and development is greatly improved, and the reliability, the readability and the portability of the program codes are enhanced. The structure of the detection node is shown in fig. 4.
3. Design of control nodes
The control node realizes information interaction with the gateway node through a self-organizing LoRa network, and comprises 4 digital-to-analog conversion circuits, an STM32 microprocessor, 4 external device controllers and a LoRa communication module SX1278, wherein the 4 digital-to-analog conversion circuits are used for controlling external devices; the 4 external equipment controllers are respectively a temperature controller, a humidity controller, a wind speed controller and a garbage cleaning device controller. The control node structure is shown in fig. 5.
4. Gateway node design
The gateway node comprises an SX1278 module, an NB-IoT module, an STM32 microprocessor and an RS232 interface, the gateway node comprises an ad hoc network for realizing communication between the SX1278 module and the detection node and the control node, the NB-IoT module realizes data bidirectional interaction between the gateway and the cloud platform, and the RS232 interface is connected with the field monitoring end to realize information interaction between the gateway and the field monitoring end. The gateway node structure is shown in fig. 6.
5. Site monitoring end software
The on-site monitoring end is an industrial control computer and mainly realizes the acquisition of environmental parameters and the control of external equipment, the information interaction with the gateway node is realized, and the on-site monitoring end mainly has the functions of communication parameter setting, data analysis and data management, environmental parameter processing and garbage cleaning subsystem. The management software selects Microsoft visual++6.0 as a development tool, and calls an Mscomm communication control of the system to design a communication program, and the function of the field monitoring end software is shown in fig. 7. The design process of the environment parameter processing and garbage cleaning subsystem is as follows:
1. parameter detection module design
The parameter detection module consists of an NARX neural network model, an Adaline neural network model, a K-means cluster classifier, a CNN convolution-LSTM neural network model, an AANN self-association neural network model of a Vague set and a beat delay line TDL;
(1) NARX neural network model design
The parameter sensor senses a time sequence parameter value of the detected environment and respectively takes the time sequence parameter value as input of a corresponding NARX neural network model and an Adaline neural network model, and a difference output by the NARX neural network model and the Adaline neural network model is taken as a parameter fluctuation value of the detected environment; the NARX neural network model is a dynamic recurrent neural network with output feedback connection, and can be equivalent to BP neural network with input delay and output-to-input delay feedback connection in topological connection, its structure is formed from input layer, delay layer, hidden layer and output layer, in which the input layer node is used for signal input, the delay layer node is used for inputting signal and input signalAnd outputting the time delay of the feedback signal, wherein the hidden layer node performs nonlinear operation on the delayed signal by using an activation function, and the output layer node is used for performing linear weighting on the hidden layer output to obtain final network output. Output h of ith hidden layer node of NARX neural network model i The method comprises the following steps:
node output O of jth output layer of NARX neural network j The method comprises the following steps:
(2) Adaline neural network model design
The parameter sensor senses a time sequence parameter value of the detected environment and respectively takes the time sequence parameter value as input of a corresponding NARX neural network model and an Adaline neural network model, and a difference output by the NARX neural network model and the Adaline neural network model is taken as a parameter fluctuation value of the detected environment; an adaptive linear unit (Adaptive Linear Element) of the Adaline neural network model is one of the early neural network models, the input signals of which can be written in the form of vectors: x (K) = [ X ] 0 (K)，x 1 (K)，…x n (K)] T Each set of input signals corresponds to a set of weight vectors expressed as: w (K) = [ K 0 (K)，k 1 (K)，…k(K)]，x 0 (K) When the bias value is equal to minus 1, the bias value of the Adaline neural network model determines the excitation or inhibition state of the neuron, and the network output can be defined according to the input vector and the weight vector of the Adaline neural network model:
in the Adaline neural network model, a special input, namely an ideal response output d (K), is sent into the Adaline neural network model, then the comparison is carried out through the output y (K) of the network, the difference value is sent into a learning algorithm mechanism, the weight vector is adjusted until the optimal weight vector is obtained, the y (K) trend is consistent with the d (K), the weight vector adjusting process is the learning process of the network, the learning algorithm is the core part of the learning process, the weight optimization searching algorithm of the Adaline neural network model adopts the least square method of the LMS algorithm, and the Adaline neural network model outputs the linear value of the detected parameter.
(3) K-means cluster classifier design
The method comprises the steps that a plurality of time sequence parameter fluctuation values and a plurality of Adaline neural network model outputs are respectively used as inputs of corresponding K-means cluster classifiers, and a plurality of types of time sequence parameter fluctuation values and Adaline neural network model outputs which are output by the 2K-means cluster classifiers are respectively used as inputs of corresponding CNN convolution-LSTM neural network models; the core idea of the K-means clustering algorithm divides n data objects into K classes, and the square sum of all data objects in each class to the clustering center points of the class is minimized, but the clustering time is relatively long, so that the efficiency of the K-means clustering classifier is reserved for realizing rapid clustering of data, and meanwhile, the application range of the K-means clustering classifier is expanded to discrete data, and the calculation process of the K-means clustering classifier is as follows:
(a) Let i=1 from the whole sample X, randomly pick K data objects in X as the initial cluster center m j (I) Where j=1, 2, …, K.
(b) Let d (i, j) represent K cluster centers m j (I) And each object X in the big data sample X i The distance between the two parts is:
searching the smallest Euclidean distance d in the Euclidean distances corresponding to all (i, j) values of d (i, j) by using a formula (4), and obtaining the value of d (i, j) in the clustering center m j (I) Identical cluster S j Storage object x in i . Let m be j (I+1) represents a new cluster center point, and the calculation formula is as follows:
n in formula (5) j Representing the number of data objects in the j-th class.
(c) Setting a judgment criterion, judging whether the criterion is met, if yes, proceeding to the next step, and if not, proceeding to the step (b).
(d) And outputting a clustering result of big data, and determining whether to terminate the loop by using a judging criterion under normal conditions, namely considering that the division is reasonable and ending the iteration when the division results obtained by the I-th iteration and the I-1 th iteration are the same.
(4) CNN convolution-LSTM neural network model design
The time sequence parameter fluctuation values of a plurality of types output by the 2K-means cluster classifiers and the output of the Adaline neural network model are respectively used as the input of a corresponding CNN convolution-LSTM neural network model; the CNN convolution-LSTM neural network model is characterized in that the output of the CNN convolution neural network is used as the input of the LSTM neural network model, the CNN convolution neural network model can directly extract the sensitive spatial features representing the time sequence input parameter information from a large number of time sequence input parameter information by automatic mining, and the CNN convolution neural network model structure mainly comprises 4 parts: (1) an Input layer (Input). The input layer is the input of the CNN convolutional neural network model, and generally, the time sequence parameters are directly input. (2) Convolutional layer (Conv). Because the dimension of the input layer is larger, the CNN convolutional neural network model is difficult to directly and comprehensively sense all time series input parameter information, the input data is required to be divided into a plurality of parts for local sensing, global information is obtained through weight sharing, meanwhile, the complexity of the structure of the CNN convolutional neural network model is reduced, the process is the main function of the convolutional layer, and the specific flow is to utilize the convolutional kernel with a specific dimension to carry out traversal and convolution operation on the time series input parameter signals with a fixed step length, so that the mining and extraction of the sensitive characteristics of the time series input parameter signals are realized. (3) Pooling layer (Pool, also called downsampling layer). Since the data sample dimensions obtained after the convolution operation remain large, the amount of data needs to be compressed andcritical information is extracted to avoid excessive model training time and overfitting, so a pooling layer follows the convolutional layer to reduce the dimensionality. Taking the peak value characteristics of the defect characteristics into consideration, adopting a maximum value pooling method to carry out downsampling. (4) And a full connection layer. After all convolution operations and pooling operations, the time series input parameter feature extraction enters a full-connection layer, each nerve layer in the layer is fully connected with all nerve cells in the previous layer, and local feature information of the time series input parameter values extracted by the convolution layer and the pooling layer is integrated. Meanwhile, in order to avoid the over-fitting phenomenon, a lost data (dropout) technology is added in the layer, an output value passing through the last layer of the full-connection layer is transmitted to the output layer, the pooling result of the last layer is connected together in an end-to-end mode to form the output layer and is used as input of an LSTM neural network model, and the LSTM neural network model introduces a mechanism of a Memory Cell and a hidden layer State to control information transmission between the hidden layers. The memory unit of an LSTM neural network has 3 Gates (Gates) computing structures, namely an Input Gate (Input Gate), a Forget Gate (force Gate) and an Output Gate (Output Gate). The input gate can control the LSTM neural network model to input new information for adding or filtering; the forgetting gate can forget to input information of the LSTM neural network model which needs to be discarded and retain the information which is useful in the past; the output gate enables the memory unit to output only the LSTM neural network model input information associated with the current time step. The 3 gate structures perform matrix multiplication, nonlinear summation and other operations in the memory unit, so that the memory is not attenuated in continuous iteration. The long-short-term memory unit (LSTM) structure unit consists of a unit (Cell), an Input Gate (Input Gate), an Output Gate (Output Gate) and a Forget Gate (Forget Gate). The LSTM neural network model is suitable for predicting the change of the input quantity of the time sequence LSTM neural network model by a long-term memory model, effectively prevents gradient disappearance during RNN training, and is a special RNN. The LSTM neural network model can learn the input dependent information of the LSTM neural network model for a long time, and meanwhile the gradient disappearance problem is avoided. LSTM on nerveA structure called a Memory Cell (Memory Cell) is added to the neural node of the hidden layer of the RNN, which is used for memorizing dynamic change information input by the LSTM neural network model in the past, and three gate (Input, forget, output) structures are added to control the use of the LSTM neural network model input history information. The time series value input as the input quantity of the detection LSTM neural network model is set as (x) 1 ，x 2 ，…，x T ) The hidden layer state is (h 1 ，h 2 ，…，h T ) Then the time t is:
i t ＝sigmoid(W hi h t-1 +W xi X t ) (6)
f t ＝sigmoid(W hf h t-1 +W hf X t ) (7)
c t ＝f t ⊙c t-1 +i t ⊙tanh(W hc h t-1 +W xc X t ) (8)
o t ＝sigmoid(W ho h t-1 +W hx X t +W co c t ) (9)
h t ＝o t ⊙tanh(c t ) (10)
wherein i is t 、f t 、O t Representing input gate, forward gate and output gate, c t Representing a cell, W h Weights representing recursive connections, W x The sigmoid and the tanh represent the weights from an input layer to an hidden layer, are two activation functions, and the LSTM neural network model is output as a nonlinear value of the parameter level of the detected region.
(5) AANN self-association neural network model design of Vague set
The output of the plurality of CNN convolution-LSTM neural network models is used as the corresponding input of the AANN self-association neural network model of the Vague set, three parameters output by the AANN self-association neural network model of the Vague set are respectively x, t and 1-f, x is the real value of the detected parameter, t is the credibility, f is the uncertainty, 1-f-t is the uncertainty, the numerical value of the Vague set formed by x, t and 1-f is [ x, (t, 1-f) ], the AANN self-association neural network model output of the Vague set is used as the input of the beat delay line TDL, and the output of the beat delay line TDL is used as the output of the parameter detection module. The AANN self-association neural network model is a feedforward self-association neural network (AANN) with a special structure, and the AANN self-association neural network model structure comprises an input layer, a certain number of hidden layers and an output layer. Firstly, compression of input data information is achieved through an input layer, a mapping layer and a bottleneck layer of input parameters, the most representative low-dimensional subspace reflecting the system structure of the input parameters is extracted from a high-dimensional parameter space of the input parameters, noise and measurement errors in the input parameter data are filtered effectively, decompression of the input parameters is achieved through the bottleneck layer, the demapping layer and an output layer, and the compressed information is restored to each parameter value, so that reconstruction of each input parameter data is achieved. In order to achieve the purpose of compressing input parameter information, the number of nodes of a bottleneck layer of an AANN self-association neural network model is obviously smaller than that of input layers, and in order to prevent simple single mapping between input and output layers forming input parameters, besides the output layer excitation function adopts a linear function, other layers adopt nonlinear excitation functions. In essence, the first layer of the hidden layer of the AANN self-association neural network model is called a mapping layer, and the node transfer function of the mapping layer may be an S-type function or other similar nonlinear functions; the second layer of the hidden layer is called a bottleneck layer, the dimension of the bottleneck layer is the smallest in the network, the transfer function of the bottleneck layer is linear or nonlinear, the bottleneck layer avoids a mapping relation of one-to-one output and input equality which is easy to realize, the bottleneck layer enables the network to encode and compress input parameter signals to obtain a related model of input data, and input parameter decoding and decompression are carried out after the bottleneck layer to generate estimated values of the input parameter signals; the third or last layer of the hidden layer is called the demapping layer, and the node transfer function of the demapping layer is a generally nonlinear sigmoid function, and the self-associative neural network is trained by using an error back propagation algorithm.
2. ANFIS neural network model design
The outputs of the 2 parameter detection modules and the fuzzy wavelet neural network model are used as corresponding inputs of the ANFIS neural network model; the ANFIS neural network model is a self-Adaptive fuzzy inference system based on a neural network, also called as a self-Adaptive fuzzy inference system (Adaptive Neuro-Fuzzy Inference System), and combines the neural network and the self-Adaptive fuzzy inference system organically, so that the advantages of the neural network and the self-Adaptive fuzzy inference system can be brought into play, and the defects of the neural network and the self-Adaptive fuzzy inference system can be made up. The fuzzy membership function and the fuzzy rule in the ANFIS neural network model are obtained through learning of known historical data of a large amount of input parameter information, and the greatest characteristic of the ANFIS neural network model is a modeling method based on data, not based on experience or intuition. The main operation steps of the ANFIS neural network model are as follows:
layer 1: blurring the input parameter information history data, each node corresponding output may be expressed as:
the formula n is the number of the membership functions input by each network, and the membership functions adopt Gaussian membership functions.
Layer 2: the rule operation is realized, the applicability of the rule is output, and the rule operation of the ANFIS neural network model adopts multiplication as follows:
layer 3: normalizing the applicability of each rule:
layer 4: the transfer function of each node is a linear function representing a local linear model, and each adaptive node i outputs:
layer 5: the single node of the layer is a fixed node, and the output of the ANFIS neural network model is calculated as follows:
the condition parameters for determining the shape of the membership function and the conclusion parameters of the inference rules in the ANFIS neural network model can be trained through a learning process. The parameters are adjusted by adopting a linear least square estimation algorithm and a gradient descent combined algorithm. Firstly, in each iteration of the ANFIS neural network model, an input signal is transmitted forward along the network until the layer 4, and a least square estimation algorithm is adopted to adjust conclusion parameters; the signal continues to pass forward along the network until the output layer. The ANFIS neural network model transmits the obtained error signal back along the network, and the condition parameters are updated by a gradient method. By adjusting given parameters in the ANFIS neural network model in the mode, the global optimal point of conclusion parameters can be obtained, so that the dimension of a search space in a gradient method can be reduced, and the convergence rate of the ANFIS neural network model parameters can be improved. And (3) inputting parameter historical data of the ANFIS neural network model, and outputting the ANFIS neural network model as an expected attitude value of the garbage disposal device.
3. Fuzzy wavelet neural network model design
The time sequence yaw angle, pitch angle and roll angle output by the MPU6050 attitude sensor are used as the input of a corresponding parameter detection module, the output of the parameter detection module is used as the input of the fuzzy wavelet neural network model to the ANFIS neural network model, and the attitude error and the error change rate of the garbage removal device output by the fuzzy wavelet neural network model are respectively used as the input of a parameter self-adjusting factor fuzzy controller and a PID controller; the fuzzy wavelet network (Fuzzy Wavelet Network, FWNN) has good intelligence, robustness, stability and index tracking rapidity, and comprises two parts: fuzzy Neural Networks (FNNs) and Wavelet Neural Networks (WNNs). The fuzzy neural network comprises 4 basic layers: the first layer is an input layer, and each input vector corresponds to a neuron; each neuron of the second layer represents a linguistic variable value; each neuron of the third layer represents a fuzzy rule; the fourth layer is the normalization layer. Meanwhile, the input of the fuzzy neural network is used as the input of the wavelet neural network, and each fuzzy rule corresponds to one wavelet network. The wavelet basis function is a wavelet basis group obtained by translating the wavelet function, so that the wavelet neural network generated by different scale functions can capture the characteristics of different time domains and frequency domains, and different fuzzy reasoning selects the corresponding wavelet network. The wavelet has the characteristic of multi-resolution analysis, if the wavelet function is used as the excitation function of the neural network neurons, the expansion and the translation of each neuron can be adjusted, the smooth function can be learned by selecting low-scale parameters, the local singular function can be learned with higher precision by improving the scale, and the ANN precision is higher than that of the same neuron number and parameter. The fuzzy wavelet network is realized by 5 basic layers of input, fuzzification, reasoning, wavelet network layer and de-fuzzification layer, and the number of the neural network nodes of each layer is n, n multiplied by M, M, M and 3 respectively. Once the number of inputs n and rules M are determined, the structure of the FWNN model is determined. Wherein the input of the fuzzy wavelet neural network is X= [ X ] 1 ，x 2 ，…x n ]，T i Is the number of wavelets corresponding to the ith rule; w (w) ik Is a weight coefficient;
is a wavelet function, +.>
The output value of the linear combination of the local model wavelet network corresponding to the rule i is:
the first layer is an input layer: each node of the layer is directly connected with each component x of the input vector j Connection is performed, and the input value X= [ X ] 1 ，x 2 ，…x n ]Pass on to the next layer; the second layer calculates membership function values corresponding to each input variable; the third layer calculates the applicability of each rule; the fourth layer is the output of the wavelet network layer and is mainly used for output compensation; the fifth layer is a control signal output layer, also called an anti-blurring layer, at which deblurring calculation is performed. The fuzzy wavelet neural network model is output as an actual value of the pose of the garbage disposal apparatus.
4. Parameter self-adjusting factor fuzzy controller design
The attitude errors and the error change rates of the garbage removal device, which are output by the ANFIS neural network model and the fuzzy wavelet neural network model, are respectively used as the inputs of a parameter self-adjusting factor fuzzy controller and a PID controller, the output of the PID controller is used as the input of an LSTM neural network model, the outputs of the LSTM neural network model and the parameter self-adjusting factor fuzzy controller are used as the corresponding inputs of an NARX neural network controller, and the output of the NARX neural network controller is respectively used as the yaw angle, pitch angle and roll angle control quantity of the garbage removal device; the parameter self-adjusting factor fuzzy controller consists of two parts of fuzzy control and integral action which are connected in parallel, the control rule of the fuzzy controller is changed by adopting the self-adjusting factor, the better control rule is used for controlling, the performance of the fuzzy controller is improved by adjusting the self-adjusting factor, when the posture grade error of the garbage disposal device is larger, the main task of the control system is to eliminate the error, and then the self-adjusting factor takes a larger value to eliminate the posture grade error of the garbage disposal device as soon as possible; when the error is smaller, the system is close to a steady state, the main control factors are that the system is stabilized as soon as possible, the ascending speed of the system is accelerated, the control effect on the change of the attitude grade error of the garbage disposal device is highlighted for reducing the overshoot of the system, and the self-adjusting factor is selected to be smaller; as the system response approaches the desired value, both may be weighted the same as the error and its variation are smaller at this time. The LSTM neural network model refers to the design process of the LSTM neural network model in the patent parameter detection module.
5. NARX neural network controller design
The output of the LSTM neural network model and the parameter self-adjusting factor fuzzy controller are used as corresponding input of the NARX neural network controller, and the NARX neural network controller output is respectively used as yaw angle, pitch angle and roll angle control quantity of the garbage disposal device; the NARX neural network controller refers to the NARX neural network model design process in the parameter detection module of this patent.
6. Design example of environmental parameter acquisition and control platform
According to the actual condition of the environment big data detection system, the system is provided with a plane arrangement installation diagram of detection nodes, gateway nodes and field monitoring ends of an environment parameter acquisition and control platform, wherein sensors of the detection nodes are uniformly arranged in all directions of the environment according to detection requirements, and the environment parameter acquisition and adjustment is realized through the system.
The technical means disclosed by the scheme of the invention is not limited to the technical means disclosed by the embodiment, and also comprises the technical scheme formed by any combination of the technical features. It should be noted that modifications and adaptations to the invention may occur to one skilled in the art without departing from the principles of the present invention and are intended to be within the scope of the present invention.
Claims (3)
1. Intelligent rubbish clearance and environmental parameter big data thing networking systems, its characterized in that: the environment parameter collection and control platform is used for detecting, adjusting and monitoring the environment parameters; the environmental parameter big data processing subsystem realizes the control of environmental parameter processing and garbage cleaning;
the environment parameter processing and garbage cleaning subsystem consists of a parameter detection module, an ANFIS neural network model, a parameter self-adjusting factor fuzzy controller, a PID controller, an LSTM neural network model, an NARX neural network controller and a fuzzy wavelet neural network model;
the method comprises the steps that multiple groups of ammonia gas, hydrogen sulfide and carbon dioxide sensor outputs are used as inputs of corresponding parameter detection modules, multiple groups of temperature, humidity and wind speed sensor outputs are used as inputs of corresponding parameter detection modules, outputs of the parameter detection modules and the fuzzy wavelet neural network model are used as corresponding inputs of an ANFIS neural network model, attitude errors and error change rates of a garbage removal device, which is output by the ANFIS neural network model and is output by the fuzzy wavelet neural network model, are respectively used as inputs of a parameter self-adjusting factor fuzzy controller and a PID controller, the PID controller output is used as an LSTM neural network model input, the outputs of the LSTM neural network model and the parameter self-adjusting factor fuzzy controller are respectively used as corresponding inputs of a NARX neural network controller, the NARX neural network controller output is respectively used as a yaw angle, a pitch angle and a roll angle control quantity of the garbage removal device, the time sequence yaw angle, the pitch angle and the roll angle of the sensor output are used as inputs of corresponding parameter detection modules, and the parameter detection module output is used as inputs of the fuzzy wavelet neural network model;
the parameter detection module consists of an NARX neural network model, an Adaline neural network model, a K-means cluster classifier, a CNN convolution-LSTM neural network model, an AANN self-association neural network model of a Vague set and a beat delay line TDL;
the method comprises the steps that a plurality of groups of parameter sensors sense time sequence parameter values of a detected environment to be respectively used as inputs of a corresponding NARX neural network model and an Adaline neural network model, differences of the NARX neural network model and the Adaline neural network model are used as fluctuation values of detected parameter levels, a plurality of time sequence parameter fluctuation values and a plurality of Adaline neural network model outputs are respectively used as inputs of a corresponding K-means cluster classifier, a plurality of types of time sequence parameter fluctuation values and Adaline neural network model outputs which are respectively used as inputs of a corresponding CNN convolution-LSTM neural network model, a plurality of CNN convolution-LSTM neural network model outputs to be used as corresponding inputs of an AANN self-association neural network model of a Vague set, three parameters of the AANN self-association neural network model of the Vague set are respectively x, t and 1-f, x is a real value of the detected parameter, t is a reliability degree, f is an uncertainty, 1-f is a reliability degree and an uncertainty degree, 1-f is a 1-tdf is a reliability degree, 1-tdf is a uncertainty degree, and 1-tdf is a delay line, and the AANN self-association neural network model of the Vague set is output as corresponding inputs.
2. The intelligent garbage collection and environmental parameter big data internet of things system according to claim 1, wherein: the environment parameter acquisition and control platform consists of a detection node, a control node, a gateway node, a field monitoring end, a cloud platform and a mobile end APP.
3. The intelligent garbage collection and environmental parameter big data internet of things system according to claim 2, wherein: the detection node collects environmental parameters, the environmental parameters are uploaded to the cloud platform through the gateway node, data provided by the cloud platform are transmitted to the mobile terminal APP, the mobile terminal APP can monitor the environmental parameters and adjust external equipment of the control node in real time through the cloud platform, the detection node and the control node are responsible for collecting environmental parameter information and controlling environmental adjustment equipment, and two-way communication of the detection node, the control node, the field monitoring terminal, the cloud platform and the mobile terminal APP is realized through the gateway node, so that environmental parameter collection, processing and environmental adjustment equipment control are realized.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210696166.6A CN114839881B (en) | 2022-06-17 | 2022-06-17 | Intelligent garbage cleaning and environmental parameter big data Internet of things system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210696166.6A CN114839881B (en) | 2022-06-17 | 2022-06-17 | Intelligent garbage cleaning and environmental parameter big data Internet of things system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114839881A CN114839881A (en) | 2022-08-02 |
| CN114839881B true CN114839881B (en) | 2023-06-30 |
Family
ID=82574104
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210696166.6A Active CN114839881B (en) | 2022-06-17 | 2022-06-17 | Intelligent garbage cleaning and environmental parameter big data Internet of things system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114839881B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115353223A (en) * | 2022-08-10 | 2022-11-18 | 盐城工学院 | Laboratory wastewater treatment integrated method and device based on ANFIS model |
| CN115511062B (en) * | 2022-10-24 | 2023-10-24 | 淮阴工学院 | Multi-parameter detection system of inspection robot |
| CN115905938B (en) * | 2022-10-24 | 2024-04-05 | 淮阴工学院 | Storage tank safety monitoring method and system based on Internet of things |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112857445A (en) * | 2021-01-13 | 2021-05-28 | 淮阴工学院 | Food safety detection system |
| CN112881601A (en) * | 2021-01-13 | 2021-06-01 | 淮阴工学院 | Moisture detection system based on cloud platform |
| CN114358244A (en) * | 2021-12-20 | 2022-04-15 | 淮阴工学院 | Pressure big data intelligent detection system based on Internet of things |
Family Cites Families (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040153227A1 (en) * | 2002-09-13 | 2004-08-05 | Takahide Hagiwara | Fuzzy controller with a reduced number of sensors |
| CN101561427B (en) * | 2009-05-15 | 2013-07-17 | 江苏大学 | Pig house environment harmful gas multi-point measurement system based on CAN field bus |
| CN101968649B (en) * | 2010-10-18 | 2012-05-30 | 淮阴工学院 | Network type control system for live pig culturing environment and intelligent environment factor control method |
| US10055434B2 (en) * | 2013-10-16 | 2018-08-21 | University Of Tennessee Research Foundation | Method and apparatus for providing random selection and long-term potentiation and depression in an artificial network |
| CN104155925B (en) * | 2014-05-20 | 2017-02-15 | 淮安信息职业技术学院 | Henhouse micro climatic environment intelligent control system based on wireless sensor network |
| WO2017218937A1 (en) * | 2016-06-16 | 2017-12-21 | Sri International | Systems and methods for optimizing operations of computing devices using deep neural networks |
| CN112013203B (en) * | 2020-07-18 | 2021-09-24 | 淮阴工学院 | Pipe network detection system based on DRNN neural network |
| CN112000008B (en) * | 2020-07-18 | 2022-04-01 | 淮阴工学院 | Pipeline monitoring system based on neural network |
| CN112183719B (en) * | 2020-09-15 | 2024-02-02 | 北京工业大学 | Intelligent detection method for total nitrogen in effluent based on multi-objective optimization-fuzzy neural network |
| CN112561146B (en) * | 2020-12-08 | 2023-04-18 | 哈尔滨工程大学 | Large-scale real-time traffic flow prediction method based on fuzzy logic and depth LSTM |
| CN112906735B (en) * | 2021-01-13 | 2023-01-13 | 淮阴工学院 | Domestic fungus environment big data detecting system |
| CN113031555B (en) * | 2021-03-17 | 2022-03-22 | 淮阴工学院 | Intelligent purification system for harmful gas in environment of livestock and poultry house |
| CN113126676B (en) * | 2021-03-17 | 2022-05-27 | 淮阴工学院 | Livestock and poultry house breeding environment parameter intelligent control system |
| CN113065749A (en) * | 2021-03-17 | 2021-07-02 | 淮阴工学院 | Building material's curing room environment intelligent system |
| CN114390376B (en) * | 2021-12-20 | 2023-04-07 | 淮阴工学院 | Fire big data remote detection and early warning system |
| CN114397043B (en) * | 2021-12-20 | 2023-04-07 | 淮阴工学院 | Multi-point temperature intelligent detection system |
-
2022
- 2022-06-17 CN CN202210696166.6A patent/CN114839881B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112857445A (en) * | 2021-01-13 | 2021-05-28 | 淮阴工学院 | Food safety detection system |
| CN112881601A (en) * | 2021-01-13 | 2021-06-01 | 淮阴工学院 | Moisture detection system based on cloud platform |
| CN114358244A (en) * | 2021-12-20 | 2022-04-15 | 淮阴工学院 | Pressure big data intelligent detection system based on Internet of things |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114839881A (en) | 2022-08-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114839881B (en) | Intelligent garbage cleaning and environmental parameter big data Internet of things system | |
| CN115016276B (en) | Intelligent water content adjustment and environment parameter Internet of things big data system | |
| Harris et al. | Advances in neurofuzzy algorithms for real-time modelling and control | |
| CN113031555B (en) | Intelligent purification system for harmful gas in environment of livestock and poultry house | |
| CN111461413B (en) | Highway road surface performance detecting system | |
| CN114397043B (en) | Multi-point temperature intelligent detection system | |
| CN113281465A (en) | Livestock and poultry house breeding environment harmful gas detection system | |
| CN111476278A (en) | Intelligent gas concentration detection system | |
| CN115128978A (en) | Internet of things environment big data detection and intelligent monitoring system | |
| CN112904756A (en) | Pipe network big data detection system | |
| CN111444947A (en) | Bridge settlement intelligent detection system | |
| CN112903929A (en) | Food quality detection system | |
| CN114911185A (en) | Security big data Internet of things intelligent system based on cloud platform and mobile terminal App | |
| Ao et al. | Hybrid model of air quality prediction using k-means clustering and deep neural network | |
| Ji et al. | GRU-corr neural network optimized by improved PSO algorithm for time series prediction | |
| CN114995248A (en) | Intelligent maintenance and environmental parameter big data internet of things system | |
| CN115062764B (en) | Intelligent illuminance adjustment and environmental parameter Internet of things big data system | |
| CN114970745B (en) | Intelligent security and environment big data system of Internet of things | |
| CN114390376B (en) | Fire big data remote detection and early warning system | |
| CN114358244B (en) | Big data intelligent detection system of pressure based on thing networking | |
| CN114399024B (en) | Oil gas concentration big data intelligent detection system | |
| CN114415503B (en) | Temperature big data internet of things detection and intelligent control system | |
| CN115659201A (en) | Gas concentration detection method and monitoring system for Internet of things | |
| Long et al. | Short-term wind speed prediction with ensemble algorithm | |
| CN114386672A (en) | Environment big data Internet of things intelligent detection system |
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 |