CN105092092A - WSN-based pre-installed substation temperature on-line monitoring and predicting system - Google Patents

Abstract

The invention discloses a WSN-based on-line temperature monitoring and prediction system of a pre-installed substation, including a pre-installed substation wireless temperature sensor network subsystem, a pre-installed substation remote monitoring management subsystem and a pre-installed substation temperature prediction subsystem ;The pre-installed substation wireless temperature sensor network subsystem collects the temperature data of power station equipment in real time, and then transmits the temperature data to the pre-installed substation remote monitoring management subsystem and pre-installed substation temperature prediction subsystem through WSN; pre-installed substation temperature prediction The subsystem includes chaos discrimination and phase space reconstruction unit and support vector machine training and prediction unit of particle swarm optimization. The present invention monitors and predicts the temperature of the power equipment of the prefabricated substation in real time, which is of great significance for ensuring its safe and stable operation, and has a good application prospect.

Description

A WSN-based online temperature monitoring and prediction system for prefabricated substations

technical field

The invention relates to a WSN-based online temperature monitoring and prediction system of a prefabricated substation, which belongs to the application of wireless sensor networks in the field of electric power.

Background technique

The prefabricated substation has the advantages of flexible combination, convenient transportation, and convenient installation. It has attracted more and more attention from electric workers all over the world. However, the aging of equipment and excessive load will cause the temperature of substation equipment to be too high, which will cause fires, etc. Accidents happen. Therefore, the real-time monitoring and prediction of the temperature of the power equipment in the prefabricated substation is of great significance to ensure the safe and stable operation of the prefabricated substation.

The research on the temperature monitoring of power equipment in our country has a history of several decades, and some achievements have been made in the online monitoring of equipment temperature. At present, the mainstream temperature measurement methods for power equipment are manual point-by-point measurement, infrared thermal imaging, wired online monitoring and wireless monitoring. Among them, the wireless temperature measurement method has many advantages over the traditional temperature measurement method.

Contents of the invention

Purpose of the invention: In order to overcome the deficiencies in the traditional temperature measurement methods, the present invention provides a WSN-based prefabricated substation temperature online monitoring and prediction system. The technical scheme of the present invention is as follows:

A prefabricated substation temperature online monitoring and prediction system based on WSN, including a prefabricated substation wireless temperature sensor network subsystem, a prefabricated substation remote monitoring management subsystem and a prefabricated substation temperature prediction subsystem;

The pre-installed substation wireless temperature sensor network subsystem collects temperature data of power station equipment in real time, and then transmits the temperature data to the pre-installed substation remote monitoring management subsystem and pre-installed substation temperature prediction subsystem through WSN; pre-installed substation remote monitoring management The subsystem realizes remote monitoring, display, and query of substation equipment temperature data; the pre-installed substation temperature prediction subsystem includes a chaos discrimination and phase space reconstruction unit and a particle swarm optimization support vector machine training and prediction unit, which performs chaotic analysis of equipment temperature data. After verification, the phase space is further reconstructed on the data, and then the support vector machine is trained with the reconstructed phase space data, and the temperature data of the equipment is predicted according to the trained model.

The above prefabricated substation wireless temperature sensor network subsystem includes component modules and software modules. The hardware module includes a digital temperature sensor, a single-chip microcomputer, a coordinator and a remote monitoring center computer, and the software module includes a driving acquisition sub-module for the digital temperature sensor by the single-chip microcomputer and a data transmission sub-module using the ZigBee protocol.

The above-mentioned digital temperature sensor is DS18B20 of DALLAS Semiconductor Company, the single-chip microcomputer is CC2530 of TI Company, and the ZigBee solution uses TI's protocol stack Z-Stack.

The pre-installed substation remote monitoring management subsystem includes a real-time monitoring module of equipment temperature, a historical query module of equipment temperature, a system user management module and a user password modification module;

The equipment temperature real-time monitoring module includes real-time temperature data display and curve display sub-modules;

The equipment temperature history query module includes historical temperature data display, curve display and data export sub-modules;

The system user management module includes loading, adding, editing, deleting and exporting user information sub-modules.

The working steps of the above chaos discrimination and phase space reconstruction unit are as follows:

Step (1), adopting the small quantity method to carry out the solution of the maximum Lyapunov exponent of the system;

Step (2), judging the chaos of the system according to the maximum Lyapunov exponent of the system.

Above-mentioned step (1) adopts small quantity method to carry out the solution step of the maximum Lyapunov exponent of system specifically refers to:

(6a), solve the average period of temperature time series according to photon energy method;

(6b), solve the time delay of the temperature time series according to the mutual information method;

(6c), solve the embedding dimension of temperature time series according to Cao method;

(6d), phase space reconstruction is carried out according to time delay and embedding dimension;

(6e), find the corresponding nearest neighbor point according to each reference point in the reconstructed phase space;

(6f), solving the distance after each point in the phase space and the fixed discrete time step of the adjacent point;

(6g), solving the logarithm of the above-mentioned distance and the mean value in each fixed time.

The specific steps of the support vector machine training and prediction unit of the above particle swarm optimization are as follows:

(7a), processing the data after phase space reconstruction;

(7b), adopt particle swarm optimization algorithm to optimize the penalty factor and kernel function parameter of support vector machine;

(7c), initializations such as the penalty factor of the optimized support vector machine, kernel function parameters and kernel functions are set;

(7d), constructing the optimal hyperplane;

(7e), train the support vector machine through the training sample data, and construct the prediction model;

(7f), using a prediction model to predict the temperature data.

Above-mentioned (7b) uses particle swarm optimization algorithm to optimize the penalty factor and kernel function parameter specific steps of support vector machine to refer to:

(8a), initializing the position and velocity of each particle of the particle swarm and the number N of the particle swarm;

(8b), using the pre-written fitness function, calculate the fitness value Fit[i] of each particle;

(8c), judging whether to update the individual extremum, for each particle, if its fitness value Fit[i] is smaller than the individual extremum Pbest(i), that is, Pbest(i)>Fit[i], then use Fit [i] update Pbest(i);

(8d), judging whether to update the local extremum, for each particle, if the particle’s fitness value Fit[i] is smaller than the global extremum Nbest, that is, Nbest(i)>Fit[i], then use Fit[ i] update Nbest(i);

(8e), according to the formula

Vi=w*Vi+c1*r1(Pbesti-Xi)+c2*r2(Nbesti-Xi)(1)

Xi=Xi+Vi(2)

Adjust the velocity Vi and position Xi of each particle, where Vi is the velocity of the i-th particle, Xi is the position of the i-th particle, Pbesti is the individual extremum of the i-th particle, and Nbesti is the global extremum of the entire particle swarm , w is an inertia factor, c1 and c2 are learning factors, r1 and r2 represent random numbers uniformly distributed between 0 and 1;

(8f), judge whether to end the iteration, the iteration end condition is to reach the maximum number of iterations set before or the experimental error is less than the minimum error value set before, if the iteration does not end, then return to (8b).

The software and hardware of the present invention combine real-time acquisition, monitoring, query, and system management of equipment temperature data to verify the chaotic nature of the equipment temperature data, and then further use the data to reconstruct the phase space, and use the particle swarm algorithm to optimize the penalty factor and kernel of the support vector machine. Function parameters, and then use the data after reconstructing the phase space to train the support vector machine, and predict the temperature data of the equipment according to the trained model, thereby improving the accuracy of the prediction. In order to improve the accuracy of prediction, the present invention adopts particle swarm optimization algorithm on the basis of the current research to optimize the penalty factor and kernel function parameters of the support vector machine, so as to improve the accuracy of prediction. The particle swarm optimization algorithm is a global random search algorithm of swarm intelligence. The particle swarm optimization algorithm is based on the principles of "population" and "evolution". It mainly uses the basic laws of cooperation and competition among individuals to find the optimal solution. The reason for using the particle swarm optimization algorithm is that the algorithm is easy to implement and does not require too many parameters, and has high efficiency, which is especially suitable for the optimization of target parameters. Therefore, the present invention adopts the particle swarm optimization algorithm to optimize the penalty factor and the kernel function parameters of the support vector machine, then trains the SVM with the data after the phase space is reconstructed, predicts the temperature data of the equipment according to the trained model, and optimizes the support vector machine prediction system. The mean square error is reduced to 0.0211, which is reduced by 91%, thus verifying that the improved support vector machine prediction system has higher accuracy in predicting the temperature of power equipment in prefabricated substations. Compared with the traditional substation temperature monitoring method, the pre-installed substation temperature monitoring based on WSN greatly reduces the consumption of manpower and material resources, and avoids the trouble of wiring in the wired mode. It has high temperature measurement accuracy, small size, and resistance Strong interference ability and other advantages. Using corresponding technology to predict the temperature of prefabricated substation power equipment, so as to provide early warning can effectively prevent the danger of substation equipment due to excessive temperature. Using chaotic time series and support vector machine to analyze and predict the temperature data of power equipment in prefabricated substations, compared with traditional modeling and time series prediction methods, the prediction accuracy has been greatly improved.

Description of drawings

Fig. 1 is a system structure block diagram of the present invention;

Fig. 2 is a system design structural diagram of the prefabricated substation remote monitoring management subsystem of the present invention;

Fig. 3 is the flow chart of the support vector machine prediction of particle swarm optimization of the present invention;

Fig. 4 is a prediction simulation effect diagram of the particle swarm optimization support vector machine of the present invention.

detailed description

The present invention will be further described below in conjunction with the accompanying drawings. The following examples are only used to illustrate the technical solution of the present invention more clearly, but not to limit the protection scope of the present invention.

As shown in Figure 1, a prefabricated substation temperature online monitoring and prediction system based on WSN, including prefabricated substation wireless temperature sensor network subsystem, prefabricated substation remote monitoring management subsystem and prefabricated substation temperature prediction subsystem;

The pre-installed substation wireless temperature sensor network subsystem collects temperature data of power station equipment in real time, and then transmits the temperature data to the pre-installed substation remote monitoring management subsystem and pre-installed substation temperature prediction subsystem through WSN; pre-installed substation remote monitoring management The subsystem realizes remote monitoring, display, and query of substation equipment temperature data; the pre-installed substation temperature prediction subsystem includes a chaos discrimination and phase space reconstruction unit and a particle swarm optimization support vector machine training and prediction unit, which performs chaotic analysis of equipment temperature data. After verification, the phase space is further reconstructed on the data, and then the support vector machine is trained with the reconstructed phase space data, and the temperature data of the equipment is predicted according to the trained model.

The above prefabricated substation wireless temperature sensor network subsystem includes component modules and software modules. The hardware module includes a digital temperature sensor, a single-chip microcomputer, a coordinator and a remote monitoring center computer, and the software module includes a driving acquisition sub-module for the digital temperature sensor by the single-chip microcomputer and a data transmission sub-module using the ZigBee protocol.

The above-mentioned digital temperature sensor is DS18B20 of DALLAS Semiconductor Company, the single-chip microcomputer is CC2530 of TI Company, and the ZigBee solution uses TI's protocol stack Z-Stack.

As shown in Figure 2, the pre-installed substation remote monitoring management subsystem includes a real-time monitoring module of equipment temperature, a historical query module of equipment temperature, a system user management module and a user password modification module;

The equipment temperature real-time monitoring module includes real-time temperature data display and curve display sub-modules;

The equipment temperature history query module includes historical temperature data display, curve display and data export sub-modules;

The system user management module includes loading, adding, editing, deleting and exporting user information sub-modules.

As shown in Figure 4, the working steps of the chaos discrimination and phase space reconstruction unit of the present invention are as follows:

Step (1), adopting the small quantity method to carry out the solution of the maximum Lyapunov exponent of the system;

Step (2), judging the chaos of the system according to the maximum Lyapunov exponent of the system.

Above-mentioned step (1) adopts small quantity method to carry out the solution step of the maximum Lyapunov exponent of system specifically refers to:

(6a), solve the average period of temperature time series according to photon energy method;

(6b), solve the time delay of the temperature time series according to the mutual information method;

(6c), solve the embedding dimension of temperature time series according to Cao method;

(6d), phase space reconstruction is carried out according to time delay and embedding dimension;

(6e), find the corresponding nearest neighbor point according to each reference point in the reconstructed phase space;

(6f), solving the distance after each point in the phase space and the fixed discrete time step of the adjacent point;

(6g), solving the logarithm of the above-mentioned distance and the mean value in each fixed time.

The specific steps of the support vector machine training and prediction unit of the above particle swarm optimization are as follows:

(7a), processing the data after phase space reconstruction;

(7b), adopt particle swarm optimization algorithm to optimize the penalty factor and kernel function parameter of support vector machine;

(7c), initializations such as the penalty factor of the optimized support vector machine, kernel function parameters and kernel functions are set;

(7d), constructing the optimal hyperplane;

(7e), train the support vector machine through the training sample data, and construct the prediction model;

(7f), using a prediction model to predict the temperature data.

Above-mentioned (7b) uses particle swarm optimization algorithm to optimize the penalty factor and kernel function parameter specific steps of support vector machine to refer to:

(8a), initializing the position and velocity of each particle of the particle swarm and the number N of the particle swarm;

(8b), using the pre-written fitness function, calculate the fitness value Fit[i] of each particle;

(8c), judging whether to update the individual extremum, for each particle, if its fitness value Fit[i] is smaller than the individual extremum Pbest(i), that is, Pbest(i)>Fit[i], then use Fit [i] update Pbest(i);

(8d), judging whether to update the local extremum, for each particle, if the particle’s fitness value Fit[i] is smaller than the global extremum Nbest, that is, Nbest(i)>Fit[i], then use Fit[ i] update Nbest(i);

(8e), according to the formula

Vi=w*Vi+c1*r1(Pbesti-Xi)+c2*r2(Nbesti-Xi)(1)

Xi=Xi+Vi(2)

Adjust the velocity Vi and position Xi of each particle, where Vi is the velocity of the i-th particle, Xi is the position of the i-th particle, Pbesti is the individual extremum of the i-th particle, and Nbesti is the global extremum of the entire particle swarm , w is an inertia factor, c1 and c2 are learning factors, r1 and r2 represent random numbers uniformly distributed between 0 and 1;

(8f), judge whether to end the iteration, the iteration end condition is to reach the maximum number of iterations set before or the experimental error is less than the minimum error value set before, if the iteration does not end, then return to (8b).

The above are only preferred embodiments of the present invention. It should be pointed out that for those of ordinary skill in the art, without departing from the technical principle of the present invention, some improvements and deformations can also be made, and these improvements and deformations should also be It is regarded as the protection scope of the present invention.

Claims (8)

1., based on preassembled transformer station on-line temperature monitoring and a prognoses system of WSN, it is characterized in that comprising preassembled transformer station radio temperature sensor network subsystem, preassembled transformer station remote monitoring subsystem and preassembled transformer station temperature prediction subsystem;

Preassembled transformer station radio temperature sensor network subsystem Real-time Collection power station equipment temperature data, then transmit temperature data to preassembled transformer station remote monitoring subsystem and preassembled transformer station temperature prediction subsystem by WSN; Preassembled transformer station remote monitoring subsystem realizes remote monitoring display, inquiry substation equipment temperature data; The support vector machine that preassembled transformer station temperature prediction subsystem comprises Chaos Identification and phase space reconfiguration unit and particle group optimizing is trained and predicting unit, after chaotic property checking is carried out to device temperature data, further phase space reconfiguration is carried out to data, then with the data Training Support Vector Machines after phase space reconstruction, according to the model prediction device temperature data trained.

2. a kind of preassembled transformer station on-line temperature monitoring based on WSN according to claim 1 and prognoses system, is characterized in that: preassembled transformer station radio temperature sensor network subsystem comprises part module and software module.Hardware module comprises digital temperature sensor, single-chip microcomputer, telegon and remote monitoring center computer, and software module comprises single-chip microcomputer and gathers submodule to the driving of digital temperature sensor and adopt the data transmission module of Zigbee protocol.

3. a kind of preassembled transformer station on-line temperature monitoring based on WSN according to claim 2 and prognoses system, it is characterized in that: digital temperature sensor be the DS18B20 of DALLAS semiconductor company, single-chip microcomputer is the protocol stack Z-Stack that the CC2530 of TI company, ZigBee solution selects TI.

4. a kind of preassembled transformer station on-line temperature monitoring based on WSN according to claim 1 and prognoses system, it is characterized in that: preassembled transformer station remote monitoring subsystem comprises device temperature Real-Time Monitoring module, device temperature historical query module, system user administration module and user cipher modified module;

Device temperature Real-Time Monitoring module comprises the display of real time temperature data and curve display sub-module;

Device temperature historical query module comprises historical temperature data display, curve display and statistical conversion submodule;

System user administration module comprises loading, newly-increased, and editor deletes, and derives user profile submodule.

5. a kind of preassembled transformer station on-line temperature monitoring based on WSN according to claim 1 and prognoses system, is characterized in that: Chaos Identification and phase space reconfiguration cell operation step as follows:

Step (1), employing smallest number method carry out solving of the maximum Lyapunov exponent of system;

Step (2), carry out the differentiation of chaotic systems according to the maximum Lyapunov exponent of system.

6. a kind of preassembled transformer station on-line temperature monitoring based on WSN according to claim 5 and prognoses system, is characterized in that: the solution procedure that step (1) adopts smallest number method to carry out the maximum Lyapunov exponent of system specifically refers to:

(6a) average period of temperature-time sequence, is solved according to photon energy method;

(6b) time delay of temperature-time sequence, is solved according to mutual information method;

(6c) Embedded dimensions of temperature-time sequence, is solved according to Cao method;

(6d) reconstruct of phase space, is carried out according to time delay and Embedded dimensions;

(6e), corresponding nearest neighbor point is found according to each reference point in the phase space of reconstruct;

(6f) distance after the fixing discrete time step of each point and neighbor point in phase space, is solved;

(6g) logarithm of above-mentioned distance and the average in each set time, is solved.

7. a kind of preassembled transformer station on-line temperature monitoring based on WSN according to claim 1 and prognoses system, is characterized in that: the support vector machine training of particle group optimizing is as follows with the work concrete steps of predicting unit:

(7a), the data after phase space reconfiguration are processed;

(7b) penalty factor and the kernel functional parameter of particle cluster algorithm Support Vector Machines Optimized, is adopted;

(7c), the initialization such as the penalty factor of support vector machine, kernel functional parameter and the kernel function after optimizing are arranged;

(7d), optimal hyperlane is constructed;

(7e), by training sample data support vector machine is trained, structure forecast model;

(7f), forecast model is utilized to predict temperature data.

8. a kind of preassembled transformer station on-line temperature monitoring based on WSN according to claim 7 and prognoses system, is characterized in that: (7b) adopts the penalty factor of particle cluster algorithm Support Vector Machines Optimized and kernel functional parameter concrete steps to refer to:

(8a) position of each particle of, initialization population and speed and particle colony number N;

(8b), utilize the fitness function write in advance, calculate the fitness value Fit [i] of each particle;

(8c), judge whether to upgrade individual extreme value, concerning each particle, if its fitness value Fit [i] is less than individual extreme value Pbest (i), i.e. Pbest (i) >Fit [i], then use Fit [i] to upgrade Pbest (i);

(8d), judge whether to upgrade local extremum, concerning each particle, if the fitness value Fit [i] of this particle is less than global extremum Nbest, i.e. Nbest (i) >Fit [i], then use Fit [i] to upgrade Nbest (i);

(8e), according to formula

Vi＝w*Vi+c1*r1(Pbesti-Xi)+c2*r2(Nbesti-Xi)(1)

Xi＝Xi+Vi(2)

Adjust speed Vi and the position Xi of each particle, in formula, Vi is the speed of i-th particle, Xi is the position of i-th particle, Pbesti is the individual extreme value of i-th particle, Nbesti is the global extremum of whole population, w is inertial factor, c1 and c2 is Studying factors, r1 and r2 represents equally distributed random number between 0 and 1;

(8f), judge whether finishing iteration, iteration termination condition be reach before the maximum iteration time of setting or experimental error be less than before the minimum error values of setting, if not finishing iteration, then return (8b).