Remote monitoring method and system of skin effect electric heating system based on big data

Abstract

The present invention relates to the technical field of control systems, and discloses a remote monitoring method and system for a skin effect electric heating system based on big data. The present invention first introduces a linear auxiliary model to construct a minimization criterion function, and optimizes the parameters in the minimization criterion function based on an artificial rabbit optimization algorithm to obtain a skin effect electric heating system model; then, by training a fuzzy RBF neural network, the output control quantity controls the duty cycle and frequency and feeds back to a PID controller to regulate the temperature of the skin effect electric heating system; then, based on a graph neural network, time and space extraction is performed to obtain a multi-time scale time feature set and a multi-time scale space feature set, respectively, and temperature prediction is achieved based on a Bi‑GRU neural network; finally, a monitoring platform and a database are established to achieve data monitoring and storage of the skin effect electric heating system. The present invention achieves the purpose of remote monitoring by regulating the temperature of the skin effect electric heating system, and the remote monitoring results are accurate and objective.

Description

Technical Field

[0001] The present invention relates to the technical field of control systems, and in particular to a remote monitoring method and system for a skin effect electric heating system based on big data. Background Art

[0002] Skin effect electric heating technology is currently an effective means to achieve pipeline heating, but the traditional steam heating method has problems such as low thermal efficiency, the heating process cannot be controlled in real time, and the heating temperature is uneven; and in the pipeline transportation process, high-tech technologies such as big data are not used, resulting in inconvenient temperature monitoring process and the inability to remotely monitor temperature in real time, which is not conducive to the efficient and safe implementation of pipeline transportation. Summary of the invention

[0003] In response to the problems in the related art, the present invention provides a remote monitoring method and system for a skin effect electric heating system based on big data to overcome the above-mentioned technical problems existing in the existing related art.

[0004] To solve the above technical problems, the present invention is achieved through the following technical solutions:

[0005] The present invention is a remote monitoring method of a skin effect electric heating system based on big data, comprising the following steps:

[0006] S1. The skin effect electric heating system model includes intermediate unmeasurable variables. A linear auxiliary model is introduced to construct a minimization criterion function. For the parameters in the minimization criterion function, the optimal parameters are optimized based on the artificial rabbit optimization algorithm, and then the intermediate unmeasurable variables are inversely solved to obtain the skin effect electric heating system model;

[0007] S2. Use PID control to regulate the temperature of the skin effect electric heating system. Train the fuzzy RBF neural network based on the fuzzy neural network to obtain a fuzzy RBF neural network model. Output the control quantity of the fuzzy RBF neural network model to control the duty cycle and frequency, adjust the heating power of the skin effect electric heating system, realize temperature control, and obtain the temperature of the skin effect electric heating system.

[0008] S3. Perform convolution processing on the temperature of the skin effect electric heating system to obtain a multi-time scale feature set, perform time and space extraction on the multi-time scale feature set based on the graph neural network, and obtain a multi-time scale time feature set and a multi-time scale space feature set respectively. Temperature prediction is achieved based on the Bi-GRU neural network, and anomaly scores are introduced to determine whether the temperature data is abnormal.

[0009] S4. Establish a monitoring platform and database to monitor and store skin effect electric heating system data and conduct maintenance on abnormalities.

[0010] The invention firstly constructs a skin effect electric heating system model and introduces intermediate unmeasurable variables, obtains a minimization criterion function by using a linear auxiliary model, optimizes the parameters in the minimization criterion function by using an artificial rabbit optimization algorithm, and inversely solves the intermediate unmeasurable variables to obtain a skin effect electric heating system model; the algorithm has a strong optimization performance by using a circuitous foraging and random hiding strategy; secondly, according to the skin effect electric heating system model and temperature value, PID control is used to regulate the temperature of the skin effect electric heating system, and a fuzzy RBF neural network is obtained by combining fuzzy reasoning and neural networks, and the relevant data is fuzzified and input into the neural network, and the control quantity output by the neural network is fed back to the PID controller to realize real-time temperature control. The neural network reasoning It has fast speed, high accuracy and autonomous learning ability, and can process fuzzy information that other neural networks cannot process; after fuzzification processing, the fuzzy RBF neural network is trained, and the output control quantity controls the duty cycle and frequency, adjusts the heating power of the skin effect electric heating system, and realizes temperature control; then the temperature of the skin effect electric heating system is convolved, and the time series and spatial series are extracted in a graph learning manner to obtain multi-time scale time features and multi-time scale space features, and then the multi-time scale time features and multi-time scale space features are integrated to obtain the final input data features, and the Bi-GRU neural network is trained to realize temperature prediction. This processing method has strong robustness; finally, a monitoring platform and database are established to realize data monitoring and storage of skin effect electric heating system.

[0011] Preferably, the S1 comprises the following steps:

[0012] S11. Set the skin effect electric heating system model to A. The output of the skin effect electric heating system model is , the noise disturbance of the skin effect electric heating system model is The intermediate unmeasurable variables of the skin effect electric heating system model are and , the undetermined constant is α, then the calculation formula for the output and intermediate unmeasurable variables of the skin effect electric heating system model is as follows:

[0013] , ;

[0014] ;

[0015] in, Indicates the frequency of the current input to the skin effect electric heating system. Indicates the maximum input current frequency of the skin effect electric heating system;

[0016] In the skin effect electric heating system, there is an operator, denoted as , the polynomial of the unit shift operator η is set to and , the delay coefficient of the skin effect electric heating system is β, then the skin effect electric heating system model is expressed as follows:

[0017] ;

[0018] S12. When the skin effect electric heating system is initially operated, the output of the skin effect electric heating system model is set to 0, and the current frequency of the skin effect electric heating system input is set to 0. The output and current frequency of the skin effect electric heating system model are continuously collected after the skin effect electric heating system is operated. The intermediate unmeasurable variables in the skin effect electric heating system are solved by the linear auxiliary model of the Hammerstein module. The information vector of the linear auxiliary model is set to , the linear auxiliary model parameter vector is , the linear auxiliary model output is , independent random white noise is , then the calculation formula of the minimization criterion function of the skin effect electric heating system at time t is as follows:

[0019] ,

[0020] Where B represents the minimization criterion function of the skin effect electric heating system, represents the set of linear auxiliary model parameters, represents the output of the skin effect electric heating system model at time t, It represents the current frequency of the skin effect electric heating system input at time t, represents the linear auxiliary model information vector at time t, represents the intermediate unmeasurable variable of the skin effect electric heating system model at time t-1, represents the output of the linear auxiliary model at time t-1, represents independent random white noise at time t-1, i1=1,2,3,…,t;

[0021] The model parameters in the linear auxiliary model parameter set are optimized so that the model parameters in the linear auxiliary model parameter set satisfy the minimization criterion function of the skin effect electric heating system at time t. The specific steps are as follows:

[0022] S121, set the fitness function as the minimization criterion function of the skin effect electric heating system at time t, set the number of artificial rabbit populations to i, the dimension of artificial rabbit individuals to j, and generate the artificial rabbit population matrix C as follows:

[0023] ,

[0024] Among them, bij represents the i-th artificial rabbit individual with dimension j in the artificial rabbit population;

[0025] In the foraging stage of the artificial rabbit population, the current number of iterations is set to d, the maximum number of iterations is set to D, and the initial position of the mth artificial rabbit individual in the artificial rabbit population at the dth iteration is , the initial position of the nth artificial rabbit individual in the artificial rabbit population at the dth iteration is ,round means rounding, e means constant, e1 is a random number and , e2 is a random number and , e3 is a random number that follows a normal distribution, represents the mapping vector, δ represents the artificial rabbit running operator, represents the running length of the artificial rabbit, and the position of the mth artificial rabbit individual in the artificial rabbit population at the d+1th iteration is recorded as , the calculation formula is as follows,

[0026] ,

[0027] , ;

[0028] S122, the artificial rabbit population is in the hiding stage, and it continuously searches for the best fitness function value. It is assumed that there are e1 holes around the mth artificial rabbit in the artificial rabbit population at the dth iteration, ε represents the hiding parameter, and ε decreases linearly from 1 to 1 / D, e4 represents the random permutation number of [1,j], e5 is a random number and , then the hiding position of the mth artificial rabbit individual in the artificial rabbit population at the dth iteration is The calculation formula is as follows,

[0029] , ;

[0030] The mth artificial rabbit in the artificial rabbit population refuses to randomly choose a cave to hide in. At this time, the mth artificial rabbit in the artificial rabbit population updates its hiding position at the d+1th iteration. The calculation formula is as follows,

[0031] ;

[0032] In the continuous iteration process, the fitness function values ​​at the dth iteration and the d+1th iteration are compared, and the best fitness function value is selected for replacement. At this time, the mth artificial rabbit individual in the artificial rabbit population chooses the cave with the best fitness function value to move. The latest position of the mth artificial rabbit individual in the artificial rabbit population at the d+1th iteration is The calculation formula is as follows,

[0033] ;

[0034] Among them, f1 represents the fitness function value at the d+1th iteration, and f2 represents the fitness function value at the dth iteration;

[0035] S123, the artificial rabbits in the artificial rabbit population will constantly change their foraging and hiding states, and an energy factor is introduced to determine the states of the artificial rabbits in the artificial rabbit population. Set e6 as a random number and , then the energy factor The calculation formula is as follows,

[0036] ;

[0037] When the energy factor is greater than 1, the artificial rabbit population is in the foraging stage, and when the energy factor is less than or equal to 1, the artificial rabbit population is in the hiding stage; when the current number of iterations is equal to the maximum number of iterations, the iteration is stopped to obtain the final position of the mth artificial rabbit individual in the artificial rabbit population, and the rectangular coordinate value of the final position of the mth artificial rabbit individual in the artificial rabbit population corresponds to the model parameter in the linear auxiliary model parameter set;

[0038] S13, substituting the model parameters in the linear auxiliary model parameter set into the minimization criterion function of the skin effect electric heating system at the time t, η represents the unit backshift operator, and the skin effect electric heating system model A is obtained as follows:

[0039] .

[0040] The invention constructs a skin effect electric heating system model, introduces intermediate unmeasurable variables, uses a linear auxiliary model to process the intermediate unmeasurable variables, constructs a minimization criterion function, optimizes the parameters in the minimization criterion function using an artificial rabbit optimization algorithm, and inversely solves the intermediate unmeasurable variables to obtain a skin effect electric heating system model; the algorithm achieves optimization by simulating the artificial rabbit's circuitous foraging and random hiding, and has a fast convergence speed and a strong optimization capability.

[0041] Preferably, S2 comprises the following steps:

[0042] S21. Use PID (proportional integral differential) control to regulate the temperature of the skin effect electric heating system. The theoretical input value is obtained from the skin effect electric heating system model. The temperature deviation is calculated based on the actual input value and recorded as a temperature deviation set. The difference between the current actual input value and the previous actual input value is calculated and recorded as a temperature change rate set. The input set to be processed E1={(g1,h1),(g2,h2),…,(g f ,h f )}, where gf represents the fth temperature deviation, h f Represents the f-th temperature change rate. Based on the fuzzy neural network, the fuzzy RBF (radial basis function) neural network model is obtained by training the fuzzy RBF neural network. The specific steps are as follows:

[0043] S211, input the input set to be processed into the fuzzy neural network input layer, connect the nodes of the fuzzy neural network input layer and the input data to be processed of the input set to be processed to obtain the connected input set; then input the connected input set into the fuzzification layer of the fuzzy neural network, set the nodes of the fuzzification layer of the fuzzy neural network to represent the language variable values, and the membership function of the fuzzification layer of the fuzzy neural network is a Gaussian function, and obtain the basic domain of the connected input set corresponding to the temperature deviation and the temperature change rate, set the fuzzy language for the basic domain, generate a fuzzy language set, and calculate the initial parameter proportional coefficient, initial integral coefficient and initial differential coefficient of the PID controller according to the fuzzy language set; the nodes of the fuzzy reasoning layer of the fuzzy neural network correspond to fuzzy rules, set the deviation threshold to ω, and in the fuzzy rules, when the temperature deviation is greater than ω, increase the initial parameter proportional coefficient, decrease the initial integral coefficient and decrease the initial differential coefficient; when the temperature deviation is equal to ω, do not adjust the initial parameter proportional coefficient, initial integral coefficient and initial differential coefficient; when the temperature deviation is less than ω, reduce the initial parameter proportional coefficient, increase the initial integral coefficient and increase the initial differential coefficient;

[0044] S212, introducing the duty cycle and frequency of the skin effect electric heating system adjustment parameters, discretizing the connected input set, duty cycle and frequency by the fuzzy rule, outputting the fuzzy vector, and obtaining the processed input matrix as follows,

[0045] ,

[0046] in, Indicates The dimensions are The processed input data;

[0047] The processed input matrix is ​​input into the fuzzy RBF neural network, and it is continuously iterated. The current number of iterations is set to g, the maximum number of iterations is set to G, and the error threshold is set to , when the error of the fuzzy RBF neural network output result is less than Or when the current number of iterations is equal to the maximum number of iterations, stop the iteration and get the trained fuzzy RBF neural network. Otherwise, keep adjusting the membership function and weight until the error of the fuzzy RBF neural network output is less than Or the current number of iterations is equal to the maximum number of iterations; obtain a new theoretical input value and a new actual input value, calculate the detection temperature deviation set and the detection temperature change rate set, and obtain the detection input set after the detection temperature deviation set and the detection temperature change rate set are fuzzy processed. The detection input set is input into the trained fuzzy RBF neural network, and the accuracy threshold is set to , when the accuracy of the trained fuzzy RBF neural network output result is less than When , the iteration stops and the fuzzy RBF neural network model is obtained, otherwise S212 is repeated until the accuracy of the trained fuzzy RBF neural network output result is less than ;

[0048] S22, obtaining the current actual temperature value, the current theoretical temperature value, the current duty cycle and the current frequency, fuzzifying and discretizing the current actual temperature value, the current theoretical temperature value, the current duty cycle and the current frequency to obtain a verification matrix, inputting the verification matrix into the fuzzy RBF neural network model, the fuzzy RBF neural network model outputs a control quantity, and inputting the control quantity into a PID controller, and the PID controller adjusts the duty cycle and frequency by adjusting the proportional coefficient, the integral coefficient and the differential coefficient; the duty cycle and frequency control the heating power of the skin effect electric heating system, and the rated heating power upper limit threshold is set to , the lower limit threshold of rated heating power is , when the heating power of the skin effect electric heating system is less than When the heating power of the skin effect electric heating system is greater than When , reduce the heating power of the skin effect electric heating system to obtain the temperature of the skin effect electric heating system.

[0049] The invention regulates the temperature of the skin effect electric heating system by using PID control. The input data is fuzzy processed, fuzzy rules are used to assign connections, and the fuzzy neural network and RBF neural network are combined to obtain a fuzzy RBF neural network. The duty cycle and frequency are controlled by outputting the control quantity of the fuzzy RBF neural network model to adjust the heating power of the skin effect electric heating system and achieve temperature control. The neural network can effectively adjust the control quantity and achieve precise temperature control.

[0050] Preferably, S3 comprises the following steps:

[0051] S31, the temperature of the skin effect electric heating system is transmitted to the monitoring platform, and the monitoring platform monitors and records the temperature of the skin effect electric heating system in real time, obtains the temperature of the skin effect electric heating system in previous years, and obtains the temperature collection of previous years. ,in Indicates The previous years' temperature set is set to , generate the temperature matrix of previous years as follows,

[0052] ,

[0053] in, Indicates The dimensions are Temperatures in previous years;

[0054] Fill the temperature matrix of previous years with data and set the convolutional neural network convolution kernel size to , the number of convolutional neural network convolution kernels is , fill in the left and right sides of the temperature matrix of previous years The filled temperature matrix of previous years is obtained by convolving the filled temperature matrix of previous years with the convolutional neural network convolution kernel to extract time series features, and the time series features are spliced ​​to obtain temperature characteristics, the The temperature features are fused to obtain a multi-time scale feature set;

[0055] S32, the multi-time scale feature is represented according to the graph neural network to obtain a multi-time scale feature graph, and the graph neural network is set to include nodes, the lines between nodes represent adjacent edges, the activation function is the Relu function, after the multi-time scale features are processed by the activation function, the attention coefficients of the adjacent edges are obtained, and the attention coefficients of the nodes connected to the adjacent edges are summed to obtain a multi-time scale spatial feature set;

[0056] The sliding window size is set to k, the multi-time scale feature includes time nodes, and time nodes within a size of k are selected from the multi-time scale feature, recorded as nodes to be processed, and a time graph E3={E4,E5} is constructed for the nodes to be processed, where E4 represents a time graph node set, and E5 represents a time graph edge set; the first node in the time graph node set is connected to all nodes except the first node, the last node in the time graph node set is connected to all nodes except the last node, and all nodes in the time graph node set except the first node and the last node are connected to two adjacent nodes, and the nodes in the time graph are spliced ​​to obtain a multi-time scale time feature set;

[0057] The relative weight coefficient is set to γ, the multi-time scale spatial feature set is recorded as F1, the multi-time scale temporal feature set is recorded as F2, represents splicing, then the calculation formula of the multi-time scale data feature set F3 is as follows:

[0058] ;

[0059] S33, dividing the multi-time scale data feature set into a multi-time scale data feature training set and a multi-time scale data feature test set, inputting the multi-time scale data feature training set into a Bi-GRU (bidirectional gated recurrent unit) neural network input layer to a hidden layer, wherein the hidden layer includes a forward process and a directional process, and Momentary state and multi-time scale data features The state at each moment is obtained, and the positive output value is obtained. Then, according to the multi-time scale data characteristics of the reverse process Momentary state and multi-time scale data features The state at the moment, obtains the reverse output value, calculates the average of the forward output value and the reverse output value, records it as the final output value, inputs the final output value into the fully connected layer for conversion, and outputs the predicted value;

[0060] The number of multi-scale convolution kernels of the Bi-GRU neural network is set to λ1, the size of the multi-scale convolution kernel is set to λ2, the activation function is set to the Relu function, the maximum number of iterations is set to F, the current number of iterations is set to l, the multi-time scale data feature training set is input into the Bi-GRU neural network, and it is continuously iterated. When the current number of iterations is equal to the maximum number of iterations, the iteration is stopped to obtain a trained Bi-GRU neural network; then the multi-time scale data feature test set is input into the trained Bi-GRU neural network, and the error threshold is set to , when the error between the output result of the trained Bi-GRU neural network and the actual value is less than , then the final Bi-GRU neural network is obtained; otherwise, S33 is repeated until the error between the output result of the trained Bi-GRU neural network and the actual value is less than ;

[0061] S34. Get the current The temperature of the skin effect electric heating system within the time is obtained, and the current multi-time scale data feature set is input into the final Bi-GRU neural network to output the temperature prediction value, which is recorded as ; m represents the temperature of the skin effect electric heating system, represents the reconstruction temperature, then the anomaly score calculation formula is as follows:

[0062] ,

[0063] in, express Skin effect electric heating system temperature at all times, express Output temperature prediction value at all times, express The reconstruction temperature at each moment, n represents the anomaly score, μ represents the hyperparameter, ;

[0064] Set the abnormal threshold to σ. When the abnormal score is less than σ, the current There is no abnormal temperature data in the skin effect electric heating system at this moment, otherwise the current If there is abnormal temperature data in the skin effect electric heating system at a certain time, the abnormal temperature data is discarded to complete the temperature prediction of the skin effect electric heating system.

[0065] The invention obtains a multi-time scale feature set by extracting features of temperatures in previous years, and uses graph learning to capture spatial series features and time series features of the multi-time scale feature set to obtain multi-time scale temporal features and multi-time scale spatial features, which are then fused to obtain the final input data features; and the Bi-GRU neural network is trained to achieve temperature prediction, and anomaly scores are introduced to achieve anomaly detection of temperature data. This processing method is fast and breaks through spatial limitations.

[0066] Preferably, S4 comprises the following steps:

[0067] S41, establish a monitoring platform to monitor the heating power, real-time temperature, and predicted temperature of the skin effect electric heating system, and when the temperature adjustment is abnormal or the temperature data is abnormal, an alarm is issued and the skin effect electric heating system is repaired;

[0068] S42, establishing a database to store the heating power, temperature, and predicted temperature values ​​of the skin effect electric heating system. When the predicted temperature in the database does not match the real-time temperature, it is displayed in real time on the monitoring platform.

[0069] The present invention also discloses a system for a remote monitoring method of a skin effect electric heating system based on big data, which specifically includes: a skin effect electric heating system model building module, a temperature control module, a temperature prediction module, and a monitoring platform and database building module;

[0070] The skin effect electric heating system model building module is used to build a skin effect electric heating system model and use the worker rabbit optimization algorithm to optimize and obtain the best parameters;

[0071] The temperature control module is used to adjust the heating power of the skin effect electric heating system based on the output control quantity of the fuzzy RBF neural network model to achieve temperature control;

[0072] The temperature prediction module is used to extract multi-scale time features and use a Bi-GRU neural network to achieve temperature prediction;

[0073] The monitoring platform and database establishment module is used to establish a monitoring platform and a database to realize data monitoring and storage of the skin effect electric heating system.

[0074] The present invention has the following beneficial effects:

[0075] 1. The invention first introduces intermediate unmeasurable variables, uses a linear auxiliary model to process the intermediate unmeasurable variables, constructs a minimization criterion function, uses an artificial rabbit optimization algorithm to optimize the parameters in the minimization criterion function, reversely solves the intermediate unmeasurable variables, and obtains a skin effect electric heating system model; the algorithm has a strong optimization performance by using a circuitous foraging and random hiding strategy.

[0076] 2. The invention regulates the temperature of the skin effect electric heating system by using PID control, obtains a fuzzy RBF neural network by combining fuzzy reasoning and neural network, and realizes real-time temperature regulation by fuzzifying relevant data and feeding back the control quantity output by the neural network to the PID controller. The neural network has fast reasoning speed, high accuracy and autonomous learning ability, and can process fuzzy information that other neural networks cannot process. The duty cycle and frequency are controlled by outputting the control quantity to adjust the heating power of the skin effect electric heating system and realize temperature regulation.

[0077] 3. The invention extracts time series and space series by means of graph learning to obtain multi-time scale time features and multi-time scale space features, which are then integrated to obtain the final input data features, and the Bi-GRU neural network is trained to achieve temperature prediction; finally, a monitoring platform and database are established to realize data monitoring and storage of the skin effect electric heating system.

[0078] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] In order to more clearly illustrate the technical solutions of the embodiments of the invention, the drawings required for describing the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For ordinary technicians in this field, they can also obtain drawings based on these drawings without paying creative work.

[0080] Figure 1 A schematic flow chart of remote monitoring of a skin effect electric heating system by a remote monitoring system of a skin effect electric heating system based on big data provided by the present invention. DETAILED DESCRIPTION

[0081] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0082] In the description of the present invention, it is necessary to understand that the terms "opening", "upper", "lower", "top", "middle", "inside" and the like indicating orientation or positional relationship are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the components or elements referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as limiting the invention.

[0083] This implementation discloses a remote monitoring method for a skin effect electric heating system based on big data, which specifically includes the following contents:

[0084] S1. The skin effect electric heating system model includes intermediate unmeasurable variables. A linear auxiliary model is introduced to construct a minimization criterion function. For the parameters in the minimization criterion function, the optimal parameters are optimized based on the artificial rabbit optimization algorithm, and then the intermediate unmeasurable variables are inversely solved to obtain the skin effect electric heating system model;

[0085] The S1 comprises the following steps:

[0086] S11. Set the skin effect electric heating system model to A. The output of the skin effect electric heating system model is , the noise disturbance of the skin effect electric heating system model is The intermediate unmeasurable variables of the skin effect electric heating system model are and , the undetermined constant is α, then the calculation formula for the output and intermediate unmeasurable variables of the skin effect electric heating system model is as follows:

[0087] , ;

[0088] ;

[0089] in, Indicates the frequency of the current input to the skin effect electric heating system. Indicates the maximum input current frequency of the skin effect electric heating system;

[0090] In the skin effect electric heating system, there is an operator, denoted as , the polynomial of the unit shift operator η is set to and , the delay coefficient of the skin effect electric heating system is β, then the skin effect electric heating system model is expressed as follows:

[0091]

[0092] S12. When the skin effect electric heating system is initially operated, the output of the skin effect electric heating system model is set to 0, and the current frequency of the skin effect electric heating system input is set to 0. After the skin effect electric heating system is operated, the output and current frequency of the skin effect electric heating system model are continuously collected. For the intermediate unmeasurable variables in the skin effect electric heating system, the linear auxiliary model of the Hammerstein module is used to solve them. The information vector of the linear auxiliary model is set to , the linear auxiliary model parameter vector is , the linear auxiliary model output is , independent random white noise is , then the calculation formula of the minimization criterion function of the skin effect electric heating system at time t is as follows:

[0093] ;

[0094] Where B represents the minimization criterion function of the skin effect electric heating system, represents the set of linear auxiliary model parameters, represents the output of the skin effect electric heating system model at time t, It represents the current frequency of the skin effect electric heating system input at time t, represents the linear auxiliary model information vector at time t, represents the intermediate unmeasurable variable of the skin effect electric heating system model at time t-1, represents the output of the linear auxiliary model at time t-1, represents independent random white noise at time t-1, i1=1,2,3,…,t;

[0095] Optimizing the model parameters in the linear auxiliary model parameter set so that the model parameters in the linear auxiliary model parameter set satisfy the minimization criterion function of the skin effect electric heating system at time t, the specific steps are as follows;

[0096] S121, set the fitness function as the minimization criterion function of the skin effect electric heating system at time t, set the number of artificial rabbit populations to i, the dimension of artificial rabbit individuals to j, and generate the artificial rabbit population matrix C as follows:

[0097] ,

[0098] Among them, b ij represents the i-th artificial rabbit individual with dimension j in the artificial rabbit population;

[0099] Set the current number of iterations to d, the maximum number of iterations to D, the artificial rabbit population is in the foraging stage, and the initial position of the mth artificial rabbit individual in the artificial rabbit population at the dth iteration is , the position of the mth artificial rabbit individual in the artificial rabbit population at the d+1th iteration is , the initial position of the nth artificial rabbit individual in the artificial rabbit population at the dth iteration is ,round means rounding, e1 is a random number and , e2 is a random number and , e3 is a random number that follows a normal distribution, represents the mapping vector, δ represents the artificial rabbit running operator, represents the running length of the artificial rabbit, e represents a constant, then the position calculation formula of the mth artificial rabbit individual in the artificial rabbit population at the d+1th iteration is as follows:

[0100] ;

[0101] , ;

[0102] S122, the artificial rabbit population is in the hiding stage, and it continuously searches for the best fitness function value. It is assumed that there are e1 holes around the mth artificial rabbit in the artificial rabbit population at the dth iteration, ε represents the hiding parameter, and ε decreases linearly from 1 to 1 / D, e4 represents the random permutation number of [1,j], e5 is a random number and , then the hiding position of the mth artificial rabbit individual in the artificial rabbit population at the dth iteration is The calculation formula is as follows,

[0103] , ;

[0104] The mth artificial rabbit in the artificial rabbit population refuses to randomly choose a cave to hide in. At this time, the mth artificial rabbit in the artificial rabbit population updates its hiding position at the d+1th iteration. The calculation formula is as follows,

[0105] ;

[0106] In the continuous iteration process, the fitness function values ​​at the dth iteration and the d+1th iteration are compared, and the best fitness function value is selected for replacement. At this time, the mth artificial rabbit individual in the artificial rabbit population chooses the cave with the best fitness function value to move. The latest position of the mth artificial rabbit individual in the artificial rabbit population at the d+1th iteration is The calculation formula is as follows,

[0107] ;

[0108] Among them, f1 represents the fitness function value at the d+1th iteration, and f2 represents the fitness function value at the dth iteration;

[0109] S123, the artificial rabbits in the artificial rabbit population will constantly change their foraging and hiding states, and an energy factor is introduced to determine the states of the artificial rabbits in the artificial rabbit population. Set e6 as a random number and , then the energy factor The calculation formula is as follows,

[0110] ;

[0111] When the energy factor is greater than 1, the artificial rabbit population is in the foraging stage, and when the energy factor is less than or equal to 1, the artificial rabbit population is in the hiding stage; when the current number of iterations is equal to the maximum number of iterations, the iteration is stopped to obtain the final position of the mth artificial rabbit individual in the artificial rabbit population, and at this time, the rectangular coordinate value of the final position of the mth artificial rabbit individual in the artificial rabbit population corresponds to the model parameter in the linear auxiliary model parameter set.

[0112] S13, substituting the model parameters in the linear auxiliary model parameter set into the minimization criterion function of the skin effect electric heating system at the time t, η represents the unit backshift operator, and the skin effect electric heating system model A is obtained as follows:

[0113]

[0114] S2. Use PID control to regulate the temperature of the skin effect electric heating system. Train the fuzzy RBF neural network based on the fuzzy neural network to obtain a fuzzy RBF neural network model. Output the control quantity of the fuzzy RBF neural network model to control the duty cycle and frequency, adjust the heating power of the skin effect electric heating system, realize temperature control, and obtain the temperature of the skin effect electric heating system.

[0115] The S2 comprises the following steps:

[0116] S21. Use PID control to regulate the temperature of the skin effect electric heating system. Obtain theoretical input values ​​from the skin effect electric heating system model. Calculate temperature deviations based on actual input values, record them as temperature deviation sets. Calculate the difference between the current actual input value and the previous actual input value, record them as temperature change rate sets. Obtain the input set to be processed E1={(g1,h1),(g2,h2),…,(g f ,h f )}, where g f represents the fth temperature deviation, h fRepresents the f-th temperature change rate. Based on the fuzzy neural network, the fuzzy RBF neural network model is obtained by training the fuzzy RBF neural network. The specific steps are as follows:

[0117] S211, input the input set to be processed into the fuzzy neural network input layer, connect the nodes of the fuzzy neural network input layer and the input data to be processed of the input set to be processed to obtain the connected input set; then input the connected input set into the fuzzification layer of the fuzzy neural network, set the nodes of the fuzzification layer of the fuzzy neural network to represent the language variable values, and the membership function of the fuzzification layer of the fuzzy neural network is a Gaussian function, and obtain the basic domain after connection corresponding to the temperature deviation and the temperature change rate in the connected input set, set the fuzzy language for the basic domain, generate a fuzzy language set, and calculate the initial parameter proportional coefficient, initial integral coefficient and initial differential coefficient of the PID controller according to the fuzzy language set; the nodes of the fuzzy reasoning layer of the fuzzy neural network correspond to fuzzy rules, set the deviation threshold to ω, and in the fuzzy rules, when the temperature deviation is greater than ω, increase the initial parameter proportional coefficient, decrease the initial integral coefficient and decrease the initial differential coefficient; when the temperature deviation is equal to ω, do not adjust the initial parameter proportional coefficient, initial integral coefficient and initial differential coefficient; when the temperature deviation is less than ω, reduce the initial parameter proportional coefficient, increase the initial integral coefficient and increase the initial differential coefficient;

[0118] S212, introducing the duty cycle and frequency of the skin effect electric heating system adjustment parameters, discretizing the connected input set, duty cycle and frequency by the fuzzy rule, outputting the fuzzy vector, and obtaining the processed input matrix as follows,

[0119] ,

[0120] in, Indicates The dimensions are The processed input data;

[0121] The processed input matrix is ​​input into the fuzzy RBF neural network, and it is continuously iterated. The current number of iterations is set to g, the maximum number of iterations is set to G, and the error threshold is set to , when the error of the fuzzy RBF neural network output result is less than Or when the current number of iterations is equal to the maximum number of iterations, stop the iteration and get the trained fuzzy RBF neural network. Otherwise, keep adjusting the membership function and weight until the error of the fuzzy RBF neural network output is less than Or the current number of iterations is equal to the maximum number of iterations; obtain a new theoretical input value and a new actual input value, calculate the detection temperature deviation set and the detection temperature change rate set, and obtain the detection input set after the detection temperature deviation set and the detection temperature change rate set are fuzzy processed. The detection input set is input into the trained fuzzy RBF neural network, and the accuracy threshold is set to , when the accuracy of the trained fuzzy RBF neural network output result is less than When , the iteration stops and the fuzzy RBF neural network model is obtained, otherwise S212 is repeated until the accuracy of the trained fuzzy RBF neural network output result is less than .

[0122] S22, obtaining the current actual temperature value, the current theoretical temperature value, the current duty cycle and the current frequency, fuzzifying and discretizing the current actual temperature value, the current theoretical temperature value, the current duty cycle and the current frequency to obtain a verification matrix, inputting the verification matrix into the fuzzy RBF neural network model, the fuzzy RBF neural network model outputs a control quantity, and inputting the control quantity into a PID controller, and the PID controller adjusts the duty cycle and frequency by adjusting the proportional coefficient, the integral coefficient and the differential coefficient; the duty cycle and frequency control the heating power of the skin effect electric heating system, and the rated heating power upper limit threshold is set to , the lower limit threshold of rated heating power is , when the heating power of the skin effect electric heating system is less than When the heating power of the skin effect electric heating system is greater than When , reduce the heating power of the skin effect electric heating system to obtain the temperature of the skin effect electric heating system.

[0123] S3. Perform convolution processing on the temperature of the skin effect electric heating system to obtain a multi-time scale feature set. Perform time and space extraction on the multi-time scale feature set based on the graph neural network to obtain a multi-time scale time feature set and a multi-time scale space feature set respectively. Temperature prediction is achieved based on the Bi-GRU neural network, and anomaly scores are introduced to determine whether the temperature data is abnormal.

[0124] The S3 comprises the following steps:

[0125] S31, the temperature of the skin effect electric heating system is transmitted to the monitoring platform, and the monitoring platform monitors and records the temperature of the skin effect electric heating system in real time, obtains the temperature of the skin effect electric heating system in previous years, and obtains the temperature collection of previous years. ,in Indicates The previous years' temperature set is set to , generate the temperature matrix of previous years as follows,

[0126] ,

[0127] in, Indicates The dimensions are Temperatures in previous years;

[0128] Fill the temperature matrix of previous years with data and set the convolutional neural network convolution kernel size to , the number of convolutional neural network convolution kernels is , fill in the left and right sides of the temperature matrix of previous years The filled temperature matrix of previous years is obtained by convolving the filled temperature matrix of previous years with the convolutional neural network convolution kernel to extract time series features, and the time series features are spliced ​​to obtain temperature characteristics, the The temperature features are fused to obtain a multi-time scale feature set;

[0129] S32, the multi-time scale feature set is represented according to a graph neural network to obtain a multi-time scale feature graph, and the graph neural network is set to include nodes, the lines between nodes represent adjacent edges, the activation function is the Relu function, after the multi-time scale features are processed by the activation function, the attention coefficients of the adjacent edges are obtained, and the attention coefficients of the nodes connected to the adjacent edges are summed to obtain a multi-time scale spatial feature set;

[0130] The sliding window size is set to k, the multi-time scale feature includes time nodes, and time nodes within a size of k are selected from the multi-time scale feature, recorded as nodes to be processed, and a time graph E3={E4,E5} is constructed for the nodes to be processed, where E4 represents a time graph node set, and E5 represents a time graph edge set; the first node in the time graph node set is connected to all nodes except the first node, the last node in the time graph node set is connected to all nodes except the last node, and all nodes in the time graph node set except the first node and the last node are connected to two adjacent nodes, and the nodes in the time graph are spliced ​​to obtain a multi-time scale time feature set;

[0131] The relative weight coefficient is set to γ, the multi-time scale spatial feature set is recorded as F1, the multi-time scale temporal feature set is recorded as F2, represents splicing, then the calculation formula of the multi-time scale data feature set F3 is as follows:

[0132] ;

[0133] S33, dividing the multi-time scale data feature set into a multi-time scale data feature training set and a multi-time scale data feature test set, inputting the multi-time scale data feature training set into the Bi-GRU neural network input layer to reach the hidden layer, wherein the hidden layer includes a forward process and a directional process, and Momentary state and multi-time scale data features The state at each moment is obtained, and the positive output value is obtained. Then, according to the multi-time scale data characteristics of the reverse process Momentary state and multi-time scale data features The state at the moment, obtains the reverse output value, calculates the average of the forward output value and the reverse output value, records it as the final output value, inputs the final output value into the fully connected layer for conversion, and outputs the predicted value;

[0134] The number of multi-scale convolution kernels of the Bi-GRU neural network is set to λ1, the size of the multi-scale convolution kernel is set to λ2, the activation function is set to the Relu function, the maximum number of iterations is set to F, the current number of iterations is set to l, the multi-time scale data feature training set is input into the Bi-GRU neural network, and it is continuously iterated. When the current number of iterations is equal to the maximum number of iterations, the iteration is stopped to obtain a trained Bi-GRU neural network; then the multi-time scale data feature test set is input into the trained Bi-GRU neural network, and the error threshold is set to , when the error between the output result of the trained Bi-GRU neural network and the actual value is less than , then the final Bi-GRU neural network is obtained; otherwise, S33 is repeated until the error between the output result of the trained Bi-GRU neural network and the actual value is less than ;

[0135] S34. Get the current The temperature of the skin effect electric heating system within the time is obtained, and the current multi-time scale data feature set is input into the final Bi-GRU neural network to output the temperature prediction value, which is recorded as ; m represents the temperature of the skin effect electric heating system, represents the reconstruction temperature, then the anomaly score calculation formula is as follows:

[0136] ;

[0137] in, express Skin effect electric heating system temperature at all times, express Output temperature prediction value at all times, express The reconstruction temperature at the moment, n represents the anomaly score, μ represents the hyperparameter, ;

[0138] Set the abnormal threshold to σ. When the abnormal score is less than σ, the current There is no abnormal temperature data in the skin effect electric heating system at this moment, otherwise the current If there is abnormal temperature data in the skin effect electric heating system at a certain time, the abnormal temperature data is discarded to complete the temperature prediction of the skin effect electric heating system.

[0139] S4. Establish a monitoring platform and database to monitor and store data of the skin effect electric heating system and perform maintenance on abnormalities;

[0140] The S4 comprises the following steps:

[0141] S41, establish a monitoring platform to monitor the heating power, real-time temperature, and predicted temperature of the skin effect electric heating system, and when the temperature adjustment is abnormal or the temperature data is abnormal, an alarm is issued and the skin effect electric heating system is repaired;

[0142] S42, establishing a database to store the heating power, temperature, and predicted temperature values ​​of the skin effect electric heating system. When the predicted temperature in the database does not match the real-time temperature, it is displayed in real time on the monitoring platform.

[0143] The present invention also discloses a remote monitoring system for a skin effect electric heating system based on big data, which specifically includes: a skin effect electric heating system model building module, a temperature control module, a temperature prediction module, and a monitoring platform and database building module;

[0144] The skin effect electric heating system model building module is used to build a skin effect electric heating system model and use the worker rabbit optimization algorithm to optimize and obtain the best parameters;

[0145] The temperature control module is used to adjust the heating power of the skin effect electric heating system based on the output control quantity of the fuzzy RBF neural network model to achieve temperature control;

[0146] The temperature prediction module is used to extract multi-scale time features and use a Bi-GRU neural network to achieve temperature prediction;

[0147] The monitoring platform and database establishment module is used to establish a monitoring platform and a database to realize data monitoring and storage of the skin effect electric heating system.

[0148] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0149] The preferred embodiments of the invention disclosed above are only used to help explain the invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the invention, so that those skilled in the art can understand and use the invention well.

Claims

1. A remote monitoring method for skin effect electric heating system based on big data, characterized in that: The steps include: S1. The skin effect electric heating system model includes intermediate unmeasurable variables. A linear auxiliary model is introduced to construct a minimization criterion function. The parameters in the minimization criterion function are optimized to obtain the optimal parameters. The intermediate unmeasurable variables are then inversely solved to obtain the skin effect electric heating system model. S2. Use PID control to regulate the temperature of the skin effect electric heating system. Train the fuzzy RBF neural network based on the fuzzy neural network to obtain a fuzzy RBF neural network model. Output the control quantity of the fuzzy RBF neural network model to control the duty cycle and frequency, adjust the heating power of the skin effect electric heating system, realize temperature control, and obtain the temperature of the skin effect electric heating system. S3. Perform convolution processing on the temperature of the skin effect electric heating system to obtain a multi-time scale feature set. Perform time and space extraction on the multi-time scale feature set based on the graph neural network to obtain a multi-time scale time feature set and a multi-time scale space feature set respectively. Temperature prediction is achieved based on the Bi-GRU neural network, and anomaly scores are introduced to determine whether the temperature data is abnormal. S4. Establish a monitoring platform and database to monitor and store skin effect electric heating system data and perform maintenance on abnormalities; The S1 comprises the following steps: S11, constructing a skin effect electric heating system model including intermediate unmeasurable variables, solving the intermediate unmeasurable variables in the skin effect electric heating system using a linear auxiliary model to obtain a minimization criterion function, and optimizing the parameters of the linear auxiliary model; S12, substituting the linear auxiliary model parameters into the minimization criterion function to obtain a skin effect electric heating system model; S121, set the fitness function as the minimization criterion function of the skin effect electric heating system at time t, set the number of artificial rabbit populations to i, the dimension of artificial rabbit individuals to j, and generate the artificial rabbit population matrix C as follows: ， Among them, b ij represents the i-th artificial rabbit individual with dimension j in the artificial rabbit population; In the foraging stage of the artificial rabbit population, the current number of iterations is set to d, the maximum number of iterations is set to D, and the initial position of the mth artificial rabbit individual in the artificial rabbit population at the dth iteration is , the initial position of the nth artificial rabbit individual in the artificial rabbit population at the dth iteration is ,round means rounding, e means constant, e1 is a random number and , e2 is a random number and , e3 is a random number that follows a normal distribution, represents the mapping vector, δ represents the artificial rabbit running operator, represents the running length of the artificial rabbit, and the position of the mth artificial rabbit individual in the artificial rabbit population at the d+1th iteration is recorded as , the calculation formula is as follows, ； S122, the artificial rabbit population is in the hiding stage, and it continuously searches for the best fitness function value. It is assumed that there are e1 holes around the mth artificial rabbit in the artificial rabbit population at the dth iteration, ε represents the hiding parameter, and ε decreases linearly from 1 to 1 / D, e4 represents the random permutation number of [1,j], e5 is a random number and , then the hiding position of the mth artificial rabbit individual in the artificial rabbit population at the dth iteration is The calculation formula is as follows, ； The mth artificial rabbit in the artificial rabbit population refuses to randomly choose a cave to hide in. At this time, the mth artificial rabbit in the artificial rabbit population updates its hiding position at the d+1th iteration. The calculation formula is as follows, ； In the continuous iteration process, the fitness function values ​​at the dth iteration and the d+1th iteration are compared, and the best fitness function value is selected for replacement. At this time, the mth artificial rabbit individual in the artificial rabbit population chooses the cave with the best fitness function value to move. The latest position of the mth artificial rabbit individual in the artificial rabbit population at the d+1th iteration is The calculation formula is as follows, ； Among them, f1 represents the fitness function value at the d+1th iteration, and f2 represents the fitness function value at the dth iteration; S123, the artificial rabbits in the artificial rabbit population will constantly change their foraging and hiding states, and an energy factor is introduced to determine the states of the artificial rabbits in the artificial rabbit population. Set e6 as a random number and , then the energy factor The calculation formula is as follows, ； When the energy factor is greater than 1, the artificial rabbit population is in the foraging stage, and when the energy factor is less than or equal to 1, the artificial rabbit population is in the hiding stage; when the current number of iterations is equal to the maximum number of iterations, the iteration is stopped to obtain the final position of the mth artificial rabbit individual in the artificial rabbit population, and the rectangular coordinate value of the final position of the mth artificial rabbit individual in the artificial rabbit population corresponds to the model parameter in the linear auxiliary model parameter set; S13, substituting the model parameters in the linear auxiliary model parameter set into the minimization criterion function of the skin effect electric heating system at the time t, η represents the unit backshift operator, and the skin effect electric heating system model A is obtained as follows: ； The S3 comprises the following steps: S31, obtaining the temperature of the skin effect electric heating system in previous years, generating a temperature matrix of previous years, filling the temperature matrix of previous years with data, and extracting time series features to obtain a multi-time scale feature set; S32, representing the multi-time scale feature set according to a graph neural network to obtain a multi-time scale feature graph, and performing time-space extraction to obtain a multi-time scale time feature set and a multi-time scale space feature set respectively; S33, dividing the multi-time scale data feature set into a multi-time scale data feature training set and a multi-time scale data feature test set; setting the Bi-GRU neural network activation function to the Relu function, the maximum number of iterations to F, the current number of iterations to 1, inputting the multi-time scale data feature training set into the Bi-GRU neural network, iterating continuously, and when the current number of iterations is equal to the maximum number of iterations, stopping the iteration to obtain a trained Bi-GRU neural network; then inputting the multi-time scale data feature test set into the trained Bi-GRU neural network, setting the error threshold to , when the error between the output result of the trained Bi-GRU neural network and the actual value is less than , then the final Bi-GRU neural network is obtained; otherwise, S33 is repeated until the error between the output result of the trained Bi-GRU neural network and the actual value is less than ; S34. Get the current The temperature of the skin effect electric heating system within the time is obtained, and the current multi-time scale data feature set is input into the final Bi-GRU neural network to output the temperature prediction value, which is recorded as ; m represents the temperature of the skin effect electric heating system, represents the reconstruction temperature, then the anomaly score calculation formula is as follows: ； in, express Skin effect electric heating system temperature at all times, express Output temperature prediction value at all times, express The reconstruction temperature at each moment, n represents the anomaly score, μ represents the hyperparameter, ; Set the abnormal threshold to σ. When the abnormal score is less than σ, the current There is no abnormal temperature data in the skin effect electric heating system at this moment, otherwise the current If there is abnormal temperature data in the skin effect electric heating system at a certain time, the abnormal temperature data is discarded to complete the temperature prediction of the skin effect electric heating system.

2. The remote monitoring method of the skin effect electric heating system based on big data according to claim 1 is characterized in that: The S11 comprises the following steps: S111, set the fitness function as the minimization criterion function of the skin effect electric heating system at time t, set the number of artificial rabbit population as i, the current number of iterations as d, and the maximum number of iterations as D; when the artificial rabbit population is in the foraging stage, the initial position of the mth artificial rabbit individual in the artificial rabbit population at the dth iteration is , get the position of the mth artificial rabbit individual in the artificial rabbit population at the d+1th iteration ; The artificial rabbit population is in the hiding stage, and it is constantly looking for the best fitness function value. ε is set to represent the hiding parameter, and ε is linearly reduced from 1 to 1 / D, e4 represents the random permutation number, e5 is a random number and , δ represents the artificial rabbit running operator, and the hiding position of the mth artificial rabbit individual in the artificial rabbit population at the dth iteration The calculation formula is as follows, ； The mth artificial rabbit in the artificial rabbit population refuses to randomly choose a cave to hide in. At this time, the mth artificial rabbit in the artificial rabbit population updates its hiding position at the d+1th iteration. The calculation formula is as follows, ； In the continuous iteration process, the fitness function values ​​at the dth iteration and the d+1th iteration are compared, and the best fitness function value is selected for replacement. At this time, the mth artificial rabbit individual in the artificial rabbit population chooses the cave with the best fitness function value to move. The latest position of the mth artificial rabbit individual in the artificial rabbit population at the d+1th iteration is The calculation formula is as follows, ； Among them, f1 represents the fitness function value at the d+1th iteration, and f2 represents the fitness function value at the dth iteration; S112, the artificial rabbits in the artificial rabbit population will constantly change their foraging and hiding states, and an energy factor is introduced to determine the states of the artificial rabbits in the artificial rabbit population. Set e6 as a random number and , then the energy factor The calculation formula is as follows, ； When the energy factor is greater than 1, the artificial rabbit population is in the foraging stage, and when the energy factor is less than or equal to 1, the artificial rabbit population is in the hiding stage; when the current number of iterations is equal to the maximum number of iterations, the iteration is stopped to obtain the final position of the mth artificial rabbit individual in the artificial rabbit population, and at this time, the rectangular coordinate value of the final position of the mth artificial rabbit individual in the artificial rabbit population corresponds to the model parameter in the linear auxiliary model parameter set.

3. The remote monitoring method of the skin effect electric heating system based on big data according to claim 2 is characterized in that: The S2 comprises the following steps: S21, using PID control to control the temperature of the skin effect electric heating system, and according to the temperature deviation and the temperature change rate, training the fuzzy RBF neural network based on the fuzzy neural network to obtain a fuzzy RBF neural network model; S22. Obtain a verification matrix after fuzzy processing and discretization of the current actual temperature value, the current theoretical temperature value, the current duty cycle and the current frequency. Input the verification matrix into the fuzzy RBF neural network model. The fuzzy RBF neural network model outputs a control quantity. The control quantity is input into a PID controller. The PID controller adjusts the duty cycle and frequency by adjusting the proportional coefficient, the integral coefficient and the differential coefficient to achieve temperature control and obtain the temperature of the skin effect electric heating system.

4. The remote monitoring method of the skin effect electric heating system based on big data according to claim 3 is characterized in that: The S21 comprises the following steps: S211, inputting the input set to be processed into the fuzzy neural network input layer to obtain a connected input set, discretizing the connected input set, duty cycle and frequency, outputting a fuzzy vector, and obtaining a processed input matrix; The processed input matrix is ​​input into the fuzzy RBF neural network, and it is iterated continuously. The current number of iterations is set to g, the maximum number of iterations is set to G, and the error threshold is set to , when the error of the fuzzy RBF neural network output result is less than Or when the current number of iterations is equal to the maximum number of iterations, stop the iteration and get the trained fuzzy RBF neural network. Otherwise, keep adjusting the membership function and weight until the error of the fuzzy RBF neural network output is less than Or the current number of iterations is equal to the maximum number of iterations; S212, obtain new theoretical input values ​​and new actual input values, calculate and obtain a detection temperature deviation set and a detection temperature change rate set, and obtain a detection input set after fuzzification processing of the detection temperature deviation set and the detection temperature change rate set. Input the detection input set into the trained fuzzy RBF neural network, and set the accuracy threshold to , when the accuracy of the trained fuzzy RBF neural network output result is less than When , the iteration stops and the fuzzy RBF neural network model is obtained, otherwise S212 is repeated until the accuracy of the trained fuzzy RBF neural network output result is less than .

5. The remote monitoring method of the skin effect electric heating system based on big data according to claim 4 is characterized in that: The S4 comprises the following steps: S41. Establish a monitoring platform to monitor the heating power, real-time temperature and predicted temperature of the skin effect electric heating system, and establish a database to store the heating power, temperature and predicted temperature values ​​of the skin effect electric heating system.

6. A system for implementing the remote monitoring method of the skin effect electric heating system based on big data as described in any one of claims 1 to 5, characterized in that: Specifically include: Skin effect electric heating system model building module, temperature control module, temperature prediction module, monitoring platform and database building module; The skin effect electric heating system model building module is used to build a skin effect electric heating system model and use the worker rabbit optimization algorithm to optimize and obtain the best parameters; The temperature control module is used to adjust the heating power of the skin effect electric heating system based on the output control quantity of the fuzzy RBF neural network model to achieve temperature control; The temperature prediction module is used to extract multi-scale time features and use a Bi-GRU neural network to achieve temperature prediction; The monitoring platform and database establishment module is used to establish a monitoring platform and a database to realize data monitoring and storage of the skin effect electric heating system.

Images