60results about How to "Safe and Stable Economic Operation" patented technology

The present invention discloses a regional complex distribution network dispatch control management system based on a virtual power plant, and belongs to the technical field of dispatch control management. Aiming at the problem that distribution network control becomes increasingly complex due to increasing complexity of the structure in a distribution network, sharply increased number of dispersed and distributed power supplies, increasingly growing capacity / power ratio and complex and varied regional distribution networks, the regional complex distribution network dispatch control managementsystem based on a virtual power plant is developed. The system mainly includes the construction of different types of virtual power plants and a big data cloud computing platform at a regional leveldistribution network dispatch center. By developing the dispatch control management system for a regional power grid, a complex distribution network can be connected in the form of a virtual power plant for unified coordination and optimization management, and the system is a new type of complex distribution network management mode. The regional complex distribution network dispatch control management system can be widely applied to distributed new energy-rich areas, and is of practical significance for developing virtual power plants, developing renewable green energy and achieving energy transformation.

The invention discloses a method for implementing a power system stabilizer. The method includes steps of acquiring a rotation speed variable quantity via a blocking filter link according to the rotation speed and active power of a rotor of a power generator, and respectively acquiring low-band differential signals, medium-band differential signals and high-band differential signals via a differential filter; parallelly adding the low-band differential signals, the medium-band differential signals and the high-band differential signals together to acquire rotation speed variation comprehensive signals of the power generator; measuring phase-shift characteristics of the rotation speed variation comprehensive signals and the rotation speed and uncompensated phase-shift characteristics of an excitation control system of the power generator by a forced oscillation process, computing required phase-shift characteristics among the rotation speed variation comprehensive signals and the power system stabilizer, designing multi-order lead-lag link phase-shift parameters which are serially multiplied by one another, performing phase compensation to meet the required phase-shift characteristics; finally outputting signals of the power system stabilizer and superposing the signals of the power system stabilizer on a reference voltage of an excitation regulator of the power generator. The method has the advantage that active power low-frequency oscillation of the synchronous generator can be effectively suppressed.

The invention discloses a multi-component composite energy storage system grid combination control method based on power prediction, comprising steps of calculating difference value PV-L of the photovoltaic output power prediction value PV and the load power prediction value PL according to the photovoltaic output power prediction value PV and the load power prediction value PL and, utilizing a wavelet packet composition method to map PV-L to the subspaces of m wavelet packets to obtain signals with different frequencies, using the frequency signals which are close to the responding frequency of a storage battery in the frequency signals as low frequency signals P low which are complimented by the storage battery and using the rest of the frequency signals and the frequency signals which are not absorbed by the battery as high frequency signals P high which are complimented by a super capacitor, using an external loop control method and an inner loop current control method to output the real power of the storage battery and the super capacitor and control the charging and discharging power of the storage battery and the super capacitor by utilizing the state of charges of the super capacitor and the current, the state of charge and maximum charge-discharge power of the storage battery according to the real condition. The advantages of the invention are safe, stable, and economic in operation. Furthermore, the times of charging and discharging of the storage battery are reduced.

The invention discloses a boiler main steam temperature control method with frequent fluctuation of an AGC load instruction of a power grid, and belongs to the technical field of boiler main steam temperature control. According to the method, the variation trend of the main steam temperature of a boiler and the difference between the measured value of the main steam temperature of the boiler and the set value of the main steam temperature are comprehensively judged and controlled in advance through a main steam temperature super-differential feedforward control loop; and the desuperheating water jet control logic circuit for preventing the main steam temperature from overshooting is used for working conditions of frequent fluctuation of the AGC load instruction of the power grid, and the main steam temperature is controlled within a set range. According to the method, the main steam temperature regulation quality can be greatly improved, the fluctuation range of the main steam temperature of the boiler is reduced, and the unit efficiency is improved; after using this control scheme, the main steam temperature fluctuation range is reduced by 3-5 DEG C than before; and with the continuous development and maturity of the main steam temperature control technology, the foundation can be laid for improving the safe, stable and efficient economic operation of the unit and achieving the long-term goal of energy-saving optimization.

The present invention provides an optimal power flow calculating method for equivalent interconnected power network on the basis of consistency of power flow, sensitivity and constraint. The method includes the steps as follows: firstly, performing a calculation by utilizing the optimal power flow calculating method so as to acquire the available capacity of the outside network before equivalent; secondly, keeping the constant available capacity before equivalent and after equivalent and the constant constraint information on the basis of consistency satisfaction of power flow and sensitivity, and deducing the constraint condition of the equivalent network; and finally, establishing a new optimal power flow model on the basis of the network after equivalent and the constrain condition. The optimal power flow calculating method calculates the available capacity of boundary nodes and boundary sections of the outside network and deduces the equivalent constraint condition on the basis of consistency of the available capacity before equivalent and after equivalent so that the accuracy of the equivalent constraint calculating is increased. The calculating accuracy of the method provided by the invention is higher than the current optimal power flow calculating method without regard to constraint, and the method can well simulate the running condition of the actual outside network and provide the internal network with suitable power supporting so as to ensure the safe, stable and economic running of the interconnected power network.

The invention provides a multi-target voltage optimizing method of a wind power plant cluster. The wind power plant cluster is controlled as a whole, the minimum wind power plant grid-connection point voltage deviation, the minimum wind power plant cluster area transmission loss and the maximum wind power plant cluster voltage stabilization margin serve as the target functions, a wind power plant reactive power adjustable capacity constraint and wind power plant grid-connection point voltage upper and lower limit constraints are taken into account, the voltage optimized value of each wind power plant grid-connection point is obtained through optimizing calculation, and optimal control over the voltage of the wind power plant cluster is realized. The multi-target voltage optimizing method can be integrated in a wind power plant cluster control system and provides support for safe, stable and economical operation of the wind power plant cluster.

The invention relates to a method for calibrating an on-line chromatographic monitoring device of a transformer station without shutdown. The method comprises the following steps: allowing an oil outlet and an oil return opening of a chromatographic calibrating apparatus to be communicated with an oil inlet pipe and an oil outlet pipe of the on-line monitoring device respectively so as to form a basic loop of an on-line monitoring device calibration system; preparing a standard oil sample for calibration in the chromatographic calibrating apparatus; opening the basic loop of the on-line monitoring device calibration system through a control valve and starting the on-line monitoring device for detection of the concentration of gas dissolved in the standard oil sample prepared in the chromatographic calibrating apparatus; introducing carrier gas into a chromatographic instrument for 10 min; starting the chromatographic instrument, weighing a certain amount of the standard oil sample from the chromatographic calibrating apparatus after the base line of the chromatographic instrument becomes stable, and detecting the standard oil sample by using the chromatographic instrument; and respectively recording the detection data of the on-line monitoring device and the chromatographic instrument, and on the basis of the detection data of the chromatographic instrument, analyzing whether the relative measurement errors and repeatability of the on-line monitoring device meets requirements.

The invention discloses an ultra-short-term wind power prediction method based on an autoregression moving average model. The ultra-short-term wind power prediction method based on the autoregression moving average model comprises the steps that data are input to enable parameters of the autoregression moving average model to be obtained; input data required by wind power prediction are input into the autoregression moving average model determined according to the parameters of the autoregression moving average model, so that a prediction result is obtained. Key information is provided for new energy power generation real-time scheduling, a new energy power generation day-ahead plan, a new energy power generation monthly plan, new energy power generation capability evaluation and wind curtailment power estimation by predicting the wind power generated during wind power generation. The ultra-short-term wind power prediction accuracy is effectively improved due to the fact a composite data source is introduced, and thus the on-grid energy of new energy resources is effectively increased on the premise that safe, stable and economical operation of a power grid is guaranteed.

The invention relates to a steam turbine cylinder joint surface deformation processing method. The method comprises following steps:1) sending a low-pressure inner upper cylinder to a steam turbine factory for machining and rubbing down; 2) polishing the lower cylinder joint surface of a low-pressure inner lower cylinder based on the machined inner upper half cylinder; 3) cleaning burrs on the machined cylinder surface and traces remained by the machining; 4) hoisting all carrier rings and separator plates in the lower cylinder; 5) cleaning the surface of the lower cylinder; 6) making analysis according to measured results and traces pressed on the lower cylinder joint surface; 7) employing a mechanical polishing tool to polish the lower cylinder joint surface from the thickest position to be polished and gradually and radially enlarging the polishing area, wherein the joint surface is repeatedly polished in the vertical and horizontal directions by a long flat ruler during each polishing process; 8) finely polishing the joint surface when a set scraping value of a depth mark is remained; and 9) repeating the step 8) until the joint surface gap value meets requirements. Compared with the prior art, the cylinder deformation can be completely eliminated through the method, and the maintenance is convenient and reliable.

The invention discloses an electric car distributed charge-discharge scheduling policy based on three-phase load balance. The scheduling policy is characterized in that the policy is used for eliminating adverse impacts brought by disordered charge of an electric car, and the problems of large calculation amount and strict communication bandwidth requirement of centralized management can be effectively relieved. According to the policy, an electric car charge-discharge optimization model is established, valley filling at night and V2G service provision at a load peak are realized, and then smooth system loading is performed; based on this, three-phase load balance constraint is considered to ensure safe and stable operation of a system; and finally, the control policy proposed in the invention is verified through an IEEE33 node power distribution network test system, the result indicates that distributed management can achieve a centralized control effect, peak clipping and valley filing are realized, and three-phase load imbalance is reduced.

The invention discloses a photovoltaic generation power prediction method based on a self-learning radial basis function. The photovoltaic generation power prediction method based on the self-learning radial basis function comprises the steps that model training is conducted to enable an SVM model to be obtained; data required by photovoltaic generation power prediction are input into the SVM model obtained through training, so that a prediction result is obtained. Key information is provided for new energy power generation real-time scheduling, a new energy power generation plan, new energy power generation capability evaluation and wind curtailment power estimation by predicting the photovoltaic generation power generated during photovoltaic generation. The ultra-short-term prediction accuracy is effectively improved by the adoption of a composite data source, and thus high-accuracy short-term photovoltaic generation power prediction is achieved.

The invention discloses a day-ahead plan making method and system considering a gas turbine unit. The method comprises steps of obtaining a load prediction value of a power grid and prediction power of a new energy unit; substituting the load prediction value of the power grid and the prediction power of the new energy unit into a pre-constructed optimization model for calculation to obtain the output power of each unit; making a day-ahead plan of the power grid based on the output power of each unit; wherein the unit comprises a gas unit, a thermal power unit and a new energy unit. an optimization model being constructed on the basis of the load prediction value of the gas unit in multiple application modes, the power grid frequency modulation standby demand, the power grid accident standby demand and the prediction power of the new energy unit as constraints and with the minimum operation cost of each unit and the minimum electricity abandoning power of the new energy unit as the target. The method is advantaged in that a gas turbine unit participating in power grid dispatching is brought into full play, the regulation and control level and utilization efficiency of the gas turbine unit in multiple modes are improved, the wind power consumption level is improved, and safe and stable operation of a power grid is guaranteed.

The invention discloses an ultra-short-term photovoltaic generation power prediction method based on a self-learning composite data source. The ultra-short-term photovoltaic generation power prediction method based on the self-learning composite data source comprises the steps that data are input to enable parameters of an autoregression moving average model to be obtained; input data required by photovoltaic generation power prediction are input into the autoregression moving average model determined according to the parameters of the autoregression moving average model, so that a prediction result is obtained, post-evaluation is conducted on the prediction result, namely the error between a predicted value and a measured value is analyzed, and order determination of the model and estimation of the parameters of the model are conducted again if a predicted error is larger than an allowable maximum error. Key information is provided for new energy power generation real-time scheduling, a new energy power generation day-ahead plan, a new energy power generation monthly plan, new energy power generation capability evaluation and wind curtailment power estimation by predicting the photovoltaic generation power generated during photovoltaic generation. The ultra-short-term photovoltaic generation power prediction accuracy is effectively improved due to the fact the composite data source is introduced, and thus the on-grid energy of new energy resources is effectively increased on the premise that safe, stable and economical operation of a power grid is guaranteed.

The invention discloses a new stator structure of high-pressure and hyperhigh-pressure generator, which consists of stator iron core (7) overlapped by stator punching sheets and frame (4), wherein the stator winding (1) is winded by high-pressure cable with round cross section to place the :sugarcoated haws string' in the stator groove (2); one cable is pierced in each groove of the stator. The invention insulates laminate, groove and phase without turn insulating to save cost and simplify operation, which improves the power quality for wind power generating, magnetic levitation train and diving and oil-submersible motors.

The invention discloses an ultra-short-term photovoltaic generation power prediction method based on a composite data source autoregression model. The ultra-short-term photovoltaic generation power prediction method based on the composite data source autoregression model comprises the steps that data are input to enable parameters of the autoregression model to be obtained; input data required by photovoltaic generation power prediction are input into the autoregression model which is determined according to the parameters of the autoregression model, so that a prediction result is obtained; model training basic data are input, order determination is conducted on the autoregression model AR(p) according to a residual variogram method, and the parameters of the model AR(p) with the determined order are estimated according to a moment estimation method. Key information is provided for new energy power generation real-time scheduling, a new energy power generation day-ahead plan, a new energy power generation monthly plan, new energy power generation capability evaluation and wind curtailment power estimation by predicting the photovoltaic generation power generated during photovoltaic generation. The ultra-short-term photovoltaic generation power prediction accuracy is effectively improved due to the fact a composite data source is introduced, and thus the on-grid energy of new energy resources is effectively increased on the premise that safe, stable and economical operation of a power grid is guaranteed.

The invention discloses an ultra-short-term forecasting method of wind power plant generated power. The method includes the steps of A, acquiring the actually-measured power data and short-term forecasting power data of a wind power plant; B, preprocessing the acquired data; C, using a wind power plant generated power ultra-short-term forecasting model based on forecasting time duration to obtain wind power plant ultra-short-term forecasting power according to the preprocessed actually-measured power data and short-term forecasting power data; D, determining the optimal forecasting time duration of the wind power plant generated power ultra-short-term forecasting model according to the wind power plant ultra-short-term forecasting power; E, by the wind power plant generated power ultra-short-term forecasting model, performing wind power plant ultra-short-term power forecasting. By the method, the real-time scheduling plan of a power grid is optimized, coordination of conventional-energy and wind power generation is comprehensively arranged, spinning reserve capacity of a power system is reduced, operation cost is lowered, and good application prospect is achieved.

The invention relates to the technical field of power systems, in particular to a non-intrusive residential user load decomposition method based on a residual convolution module. The method comprises the following steps: acquiring training data and preprocessing the data; constructing and training a load decomposition model: inputting a total active power sequence in training data into a residual convolution module, learning active power features by taking a CNN model as a basis in the residual convolution module, adding original input data and feature data learned by the CNN through cross-layer connection, further inputting the obtained data into a GRU network to learn time sequence features, and outputting a predicted value of the active power of the target electric appliance; comparing the predicted value of the active power of the target electric appliance with a true value, and continuously adjusting network parameters of the load decomposition model to obtain a trained load decomposition model; and decomposing the total active power of the user to be decomposed through the trained load decomposition model to obtain an active power decomposition result of the target electric appliance. The method is high in decomposition precision.

The invention discloses a multi-wind-field wind speed space-time prediction method based on graph convolution and a recurrent neural network. Aiming at the problem that a convolutional neural network commonly used in an existing wind speed space-time prediction method is difficult to effectively analyze wind speed data of multiple wind fields presenting non-grid distribution in reality, a graph convolutional long-short-term memory neural network is provided to process the data. The method comprises the following steps: firstly, performing graph modeling on wind speed data of multiple wind fields based on a Pearson correlation coefficient to construct a wind speed graph signal sequence; then, using graph convolution to replace multiplication in the long short-term memory neural network, and constructing a graph convolution long short-term memory neural network; and finally, constructing a multi-wind-field wind speed space-time prediction model based on the graph convolution long-short-term memory neural network and a transfer learning principle. The space-time prediction model has good point prediction and probability prediction performance, it is verified that the accuracy of wind speed point prediction and probability prediction can be improved by fusing historical wind speed information of adjacent wind fields, and a new thought is provided for short-term wind speed prediction of multiple wind fields.

The invention discloses a small-sized relay protection intelligent terminal device adopting a clamping structure and a method for changing the relay protection device or the intelligent terminal of a switch cabinet without power outage. The device is formed by a body, a pedestal and a mounting base; wherein the pedestal is fixedly arranged on the body, all of input / output lead wires including internal AC-DC measurement, state input, control output, power supply, communication and the like access the pedestal; all of the external input / output lead wires access the mounting base; and the base and the mounting base are connected in a butt joint mode and in a splicing and clamping locked way (as figure). When the device is clamped and locked on the mounting base, the mounting process is completed, and the device can be normally operated; when the device is pulled out from the mounting base, the disassembly of the device is completed, and the device can not work; according to the invention, the mounting and maintaining work of a distribution network and a micro-grid can be simplified, the power supply reliability can be improved, and the device has positive and great significance for safety, stability and economy of power grid operation.

The invention discloses a photovoltaic generation power prediction method based on a self-learning composite data source autoregression model. The photovoltaic generation power prediction method based on the self-learning composite data source autoregression model comprises the steps that data are input to enable parameters of the autoregression model to be obtained; input data required by photovoltaic generation power prediction are input into the autoregression model determined according to the parameters of the autoregression model, so that a prediction result is obtained; post-evaluation is conducted on the prediction result, namely the error between a predicted value and a measured value is analyzed, and order determination of the autoregression model AR(p) and estimation of the parameters of the model AR(p) are conducted again if a predicted error is larger than an allowable maximum error. The ultra-short-term photovoltaic generation power prediction accuracy is effectively improved due to the fact a composite data source is introduced, and thus the on-grid energy of new energy resources is effectively increased on the premise that safe, stable and economical operation of a power grid is guaranteed.

The invention discloses an optical selection method, based on total life cycle costs, of transformer station reactive power compensation devices. The method comprises the following several steps: (1) establishing total life cycle cost models of different types of reactive compensation devices, (2) determining values of calculation parameters of various costs of the total life cycle cost models, (3) performing basic calculation and comparison of total life cycle costs on the different types of reactive compensation devices according to the calculation parameters to obtain different advantages of the different types of reactive compensation devices in total life cycle cost, (4) performing sensitivity analyses on the different types of reactive compensation devices, (5) performing benefit evaluation on the different types of reactive compensation devices, and calculating results according to the total life cycle costs to obtain a comparison of cost effectiveness on the condition that a reactive compensation equivalent benefit is considered. The invention provides the optical selection method of the transformer station reactive power compensation device, improves voltage quality to the maximum extent, reduces losses to the maximum extent, and guarantees safe, stable and economic running of a power grid.

The invention provides a large power grid quiescent voltage stability optimization decision-making method and system considering situation assessment, prediction and control can be realized on line, and the problem that the system operation state is difficult to recover due to state fluctuation and decision-making time delay is solved. For a conventional prevention and control optimization model,constraint conditions in the control model are corrected by utilizing the situation evaluation indexesto compensate decision latency, the stability level of the section is predidcted and estimated through the current section; the stability level of the next moment can be estimated according to the tide state of the current moment; when the stability margin of the real-time section is not lower than the threshold value, a corresponding regulation and control strategy is executed; therefore, early judgment and early regulation and control are realized, the situation that the system operation state is difficult to recover due to state fluctuation and decision delay is avoided, and the possible voltage instability situation in the future estimated state is effectively regulated and controlledaccording to the current power flow section information.

The invention discloses a wind power prediction method based on a self-learning composite data source autoregression model. The wind power prediction method based on the self-learning composite data source autoregression model comprises the steps that data are input to enable parameters of the autoregression model to be obtained; input data required by wind power prediction are input into the autoregression model determined according to the parameters of the autoregression model, so that a prediction result is obtained; post-evaluation is conducted on the prediction result, namely the error between a predicted value and a measured value is analyzed, and order determination of the autoregression model AR(p) and estimation of the parameters of the model AR(p) are conducted again if a predicted error is larger than an allowable maximum error. Key information is provided for new energy power generation real-time scheduling, a new energy power generation day-ahead plan, a new energy power generation monthly plan, new energy power generation capability evaluation and wind curtailment power estimation by predicting the wind power generated during wind power generation. The ultra-short-term wind power prediction accuracy is effectively improved due to the fact a composite data source is introduced, and thus the on-grid energy of new energy resources is effectively increased on the premise that safe, stable and economical operation of a power grid is guaranteed.

The invention discloses an ultra-short-term photovoltaic generation power prediction method based on an autoregression moving average model. The ultra-short-term photovoltaic generation power prediction method based on the autoregression moving average model comprises the steps that data are input to enable parameters of the autoregression moving average model to be obtained; input data required by photovoltaic generation power prediction are input into the autoregression moving average model determined according to the parameters of the autoregression moving average model, so that a prediction result is obtained. Key information is provided for new energy power generation real-time scheduling, a new energy power generation day-ahead plan, a new energy power generation monthly plan, new energy power generation capability evaluation and wind curtailment power estimation by predicting the photovoltaic generation power generated during photovoltaic generation. The ultra-short-term photovoltaic generation power prediction accuracy is effectively improved due to the fact a composite data source is introduced, and thus the on-grid energy of new energy resources is effectively increased on the premise that safe, stable and economical operation of a power grid is guaranteed.

The invention provides a CFD-based transformer respiratory system capacity matching evaluation method. The method comprises the following steps: establishing a full-size three-dimensional model of a to-be-evaluated transformer and a respiratory system thereof; then, dividing a calculation domain, then setting boundary conditions and physical parameters of materials, then setting a theoretical model and a solving algorithm, setting initial conditions, calculating heat productivity and taking the heat productivity as a heat source, and finally, performing simulation to obtain oil conservator oillevel cloud chart data under different load coefficient conditions, thereby obtaining the adaptation condition of a respiratory system and a transformer. The adaptability of the transformer and the respiratory system is rapidly and accurately calculated through the CFD technology, it is guaranteed that the capacities of the transformer and the respiratory system are well matched, serious accidents such as transformer oil overflow after operation are avoided while the capacity utilization rate of the respiratory system is increased, and safe, stable and economical operation of the transformeris guaranteed.

The invention discloses a resident load prediction method based on an LSTM-SAM model and pooling, and belongs to the technical field of power systems, and the method comprises the steps: obtaining historical load data and numerical weather forecast data of a plurality of resident users, and randomly selecting a certain user as a target user; preprocessing the data of each user by adopting two-stage feature engineering; sorting the non-target users, selecting different numbers of non-target users as interconnected users, forming different pooling combinations together with the target users, constructing a training data pool based on pooling, and retaining test set data of the target users; and inputting the training data pool and the test set data of the target user into the LSTM-SAM hybrid model, obtaining the predicted value of each load component, adding the predicted values, and outputting the day-ahead load prediction result of the target user at the to-be-predicted moment and the optimal number of pooling users. According to the method, the prediction precision of the resident load is improved, guidance is provided for system scheduling and demand response implementation, and safe, stable and economical operation of a power system is guaranteed.

The invention discloses a wind power prediction method based on a self-learning radial basis function support vector machine. The wind power prediction method based on the self-learning radial basis function support vector machine comprises the steps that model training is conducted to enable an SVM model to be obtained; data required by wind power prediction are input into the SVM model obtained through training, so that a prediction result is obtained. Key information is provided for new energy power generation real-time scheduling, a new energy power generation plan, new energy power generation capability evaluation and wind curtailment power estimation by predicting the wind power generated during wind power generation. The ultra-short-term prediction accuracy is effectively improved by the adoption of a composite data source, and thus high-accuracy short-term wind power prediction is achieved.

The invention discloses an ultra-short-term wind power prediction method based on a composite data source autoregression model. The ultra-short-term wind power prediction method based on the composite data source autoregression model comprises the steps that data are input to enable parameters of the autoregression model to be obtained; input data required by wind power prediction are input into the autoregression model which is determined according to the parameters of the autoregression model, so that a prediction result is obtained, wherein the method for obtaining the parameters of the autoregression model by inputting the data specifically comprises the steps that model training basic data are input, order determination is conducted on the autoregression model AR(p) according to a residual variogram method, and the parameters of the model AR(p) with the determined order are estimated according to a moment estimation method. Key information is provided for new energy power generation real-time scheduling, a new energy power generation day-ahead plan, a new energy power generation monthly plan, new energy power generation capability evaluation and wind curtailment power estimation by predicting the wind power generated during wind power generation. The ultra-short-term wind power prediction accuracy is effectively improved due to the fact a composite data source is introduced, and thus the on-grid energy of new energy resources is effectively increased on the premise that safe, stable and economical operation of a power grid is guaranteed.

The invention provides a solid waste pyrolysis and oil gas catalytic reforming system, and belongs to the technical field of solid waste treatment. Pyrolysis reaction is carried out in a pyrolysis cavity 6 of an external heating type pyrolysis furnace 1, obtained pyrolysis oil gas enters a catalytic reforming reactor 2 to be subjected to catalytic reforming reaction, and obtained fuel gas can be subjected to resource utilization such as internal combustion power generation, boiler power generation and production of high-added-value products; pyrolytic carbon generated by pyrolysis enters a combustion area 17 for combustion, and the generated heat can be used for supporting a catalytic reforming reaction; the flue gas generated by combustion is high-temperature flue gas, and after being subjected to dust removal treatment, the flue gas enters a heat exchange cavity 7 of the external heating type pyrolysis furnace 1 to provide heat for pyrolytic reaction; and the flue gas subjected to heat exchange enters a heat exchanger 4 to perform second heat exchange with air, and the obtained hot air enters a combustion furnace 3 to provide hot air for combustion of pyrolytic carbon.