Boiler heating surface real-time monitoring system and method based on big data

Abstract

The invention relates to a boiler heating surface real-time monitoring system and method based on big data. In order to solve the problem that in the prior art, a boiler heating surface monitoring system only monitors the current temperature of a boiler heating surface and cannot predict the change trend of the heating surface, a big data method is introduced, and the boiler heating surface real-time monitoring system and method based on big data are designed. The system comprises a unit load prediction unit, a fast Fourier transform (FFT) calculation unit, a boiler heating surface temperature response calculation unit and an inverse fast Fourier transform (IFFT) calculation unit. According to a fast Fourier transform method, the system calculates the temperature response of a boiler heating surface under different frequency unit load fluctuations. The method has high real-time performance and prediction accuracy, and is particularly suitable for the peak shaving unit with the load fluctuating randomly under the current large-scale access of new energy.

Description

Technical Field

[0001] The present invention belongs to the field of thermal energy and power engineering, and in particular relates to a real-time monitoring system and method for boiler heating surfaces based on big data. Background Art

[0002] With the substantial growth of installed capacity of new energy represented by photovoltaic and wind power in the power system, the randomness of new energy output has had a significant impact on the stability of the power system. In order to reduce the fluctuation of the power system, traditional thermal power plants need to frequently adjust the peak. High-frequency and deep peak adjustment brings many problems. For example, frequent peak adjustment causes the heating surface of the boiler to frequently and rapidly increase and decrease in temperature, the internal stress of the pipe wall changes repeatedly, and the pipe is prone to creep and oxide scale shedding. When the boiler is rapidly adjusting the peak, the flow deviation on the flue gas side and the working fluid side increases, and the heating surface is prone to overheating and oxide scale formation. The heating surface of the boiler bursts due to the above reasons, which not only causes the boiler to be forced to shut down for maintenance, causing significant economic losses, but also seriously affects the dispatching and operation of the power system. By monitoring the temperature changes of the heating surface and taking corresponding measures in time, the probability of pipe burst can be effectively reduced.

[0003] The disadvantage of the technical solution for monitoring the temperature of the boiler heating surface in the prior art is that it can only monitor the current temperature of the heating surface in real time. When the system sends out an alarm signal, the phenomenon of overheating of the heating surface has already occurred. The operator cannot predict the changing trend of the heating surface and make preparations in advance. Taking into account the above-mentioned problems existing in the prior art, it is necessary to design a system that can predict the temperature of the boiler heating surface for a period of time in the future based on local meteorological, economic and other external data by introducing big data technology, and inform the operator of possible risks in advance, so that measures can be taken in advance to prevent the occurrence of problems such as overheating and creep of the superheated surface. Summary of the invention

[0004] In view of this, it is necessary to provide a real-time monitoring system and method for boiler heating surface based on big data to solve the problems existing in the prior art.

[0005] In order to achieve the above technical objectives, the present invention adopts the following technical solutions:

[0006] The present invention provides a specific embodiment of a real-time monitoring system for a boiler heating surface based on big data, and discloses a real-time monitoring system for a boiler heating surface based on big data, comprising: a unit load prediction unit, a fast Fourier transform (FFT) calculation unit, a boiler heating surface temperature response calculation unit, and an inverse fast Fourier transform (IFFT) calculation unit.

[0007] The unit load prediction unit is used to predict the unit load change of the boiler in the future period of time according to the current data of external factors, and output the unit load change curve.

[0008] The Fast Fourier Transform (FFT) calculation unit is used to transform the unit load variation curve from the time domain to the frequency domain.

[0009] The boiler heating surface temperature response calculation unit is used to calculate the temperature response of the heating surface under load changes of different frequency units.

[0010] The inverse fast Fourier transform (IFFT) calculation unit is used to transform the temperature change response of the boiler heating surface from the frequency domain to the time domain and fit it into a continuous curve output.

[0011] The unit load prediction unit includes a data processing module and a unit load prediction module, wherein the data processing module cleans, screens, removes duplicates, formats and normalizes the input current data of external factors and then outputs the data to the unit load prediction module.

[0012] The unit load prediction module adopts BP neural network to predict the unit load, and combines LM algorithm to improve the prediction accuracy and prevent the BP neural network from falling into the local optimal point.

[0013] The BP neural network is trained through external factors and historical statistical data of unit load.

[0014] The boiler heating surface temperature response calculation unit comprises an input filtering unit, a temperature response sub-calculation unit, and a response synthesis unit. The input filtering unit filters the load prediction spectrum and inputs it to the temperature response calculation sub-unit, and the temperature response calculation sub-unit inputs the temperature response spectrum to the response synthesis unit after completing the temperature response calculation.

[0015] The temperature response sub-calculation unit is divided according to the position of the heating surface, and the temperature response at different positions of the heating surface is calculated respectively.

[0016] The temperature response calculation subunit calculates the temperature response using an expert system supported by big data.

[0017] The present invention also provides a specific embodiment of a real-time monitoring method for a boiler heating surface based on big data, and discloses a real-time monitoring method for a boiler heating surface based on big data.

[0018] The method for real-time monitoring of boiler heating surfaces based on big data comprises the following steps:

[0019] S-1: Analyze the factors affecting the unit load and collect, clean and complete the data to unify the data format.

[0020] S-2: Extract data features from the data and identify key features that affect unit load forecasting.

[0021] S-3: The unit load prediction unit outputs the unit load prediction curve based on key features.

[0022] S-4: The fast Fourier transform (FFT) calculation unit converts the unit load prediction curve in the time domain into the unit load prediction spectrum in the frequency domain through the discrete Fourier transform algorithm.

[0023] S-5: The boiler heating surface temperature response calculation unit calculates and outputs the periodic response of the boiler heating surface temperature to unit load fluctuations of different frequencies based on the unit load prediction spectrum, and outputs the boiler heating surface temperature response spectrum.

[0024] S-6: The boiler heating surface temperature response spectrum is converted into a heating surface temperature response discrete curve in the time domain through the inverse fast Fourier transform IFFT calculation unit.

[0025] S-7: After the inverse transformation, the time-domain discrete curve of the temperature response of the heated surface is fitted into a continuous curve through the fitting algorithm and output.

[0026] In the S-1 step, firstly, the external factors affecting the unit load are preliminarily selected, including short-term factors such as time, temperature, electricity price, light, wind, humidity, rainfall, snowfall, holidays, etc.; long-term factors include seasonal factors, regional population, urbanization rate, GDP, proportion of primary / secondary and tertiary industries, etc., and then the data of these external factors are collected, and the data are cleaned, screened, deduplicated and formatted. Specifically, in the formatting step, the dependent variable, i.e., the historical statistical data of the unit load yi, is normalized, and the historical statistical data x of the external factors is normalized. ij After standardization, the data y i and x ij Satisfy the following formula.

[0027]

[0028] In the S-2 step, the relationship between external factors and unit load is analyzed by LASSO algorithm, and factors with poor correlation are removed. Finally, six variables including time, electricity price, temperature, seasonal factors, population, and GDP are selected as external factors and input into the unit load prediction unit.

[0029] The S-5 step can be further divided into the following steps:

[0030] The S-5.1 filter unit is used to filter the input unit load forecast curve.

[0031] S-5.2 Temperature response calculation subunit calculates temperature response.

[0032] S-5.3 Synthesis of temperature response spectrum of heated surface.

[0033] The temperature response calculation subunit temperature response calculation step can be further divided into the following steps:

[0034] An expert database is established based on the experimental test results of unit load and temperature response.

[0035] The unit load frequency and amplitude, as well as the initial conditions of the boiler are input into each temperature response subunit, which calculates the temperature response spectrum of each position of the heating surface under the frequency load fluctuation based on the expert system.

[0036] The temperature response spectrum is output to the response synthesis unit.

[0037] Further, in a preferred embodiment, the method for establishing an expert database according to the unit load test results is as follows:

[0038] Through the test, the response of the pipes at different positions of the boiler heating surface to load fluctuations is recorded. Specifically, before the boiler is put into operation, a unit test platform with electrical load is established, a set of initial parameters for the unit operation is selected, and the output power of the unit is adjusted through the electrical load so that the unit load changes according to the specified characteristic frequency and amplitude. After the system reaches a steady state, the temperature of each pipeline of the heating surface is measured by the sensor installed on the heating surface and the temperature response cycle curve is recorded. The recorded heating surface temperature response curve is subjected to spectrum analysis to obtain the temperature response spectrum of the heating surface pipeline.

[0039] The method in which the response synthesis unit performs response synthesis on the spectrum of the temperature response is:

[0040] According to the approximate model, the response synthesis unit calculates the temperature response T under different frequency load fluctuation excitation according to the following formula: i (ω i ) for superposition synthesis.

[0041] T=k1T1(ω1)+k2T2(ω2)+k3T3(ω3)+...+k n T n (ω n )

[0042] where k i is the fitting superposition correction coefficient of the spectrum of the temperature response corresponding to the i-th load fluctuation, and T is the final synthetic spectrum of the pipeline. The response synthesis unit can use a variety of methods to obtain the fitting superposition correction coefficient. Preferably, the fitting superposition correction is obtained by fitting through the least squares method, and the method of obtaining is as follows:

[0043] According to the measured spectrum of the temperature response of the pipeline, T=[T1,T2,...T i ], construct a sample with a total number of samples of m, and define the error function

[0044]

[0045] Let S′(k) = 0 and solve k by gradient descent method. j (j=1...n), thereby obtaining the fitting superposition correction coefficient of the response synthesis unit.

[0046] Compared with the prior art, the present invention uses fast Fourier transform FFT to decompose the unit load prediction curve into sinusoidal excitations of different frequencies in the frequency domain, and calculates and superimposes the boiler heating surface temperature response corresponding to the unit load at different frequencies through the boiler heating surface temperature response unit, thereby obtaining a complete temperature response spectrum of the heating surface, and then calculates the response of the heating surface temperature in the time domain through the fast Fourier inverse transform, accurately fitting the relationship between the randomly changing unit load and the heating surface temperature, thereby realizing the real-time prediction function of the heating temperature.

[0047] Furthermore, the present application provides a computer-readable storage medium, on which computer program instructions are stored. When the computer program instructions are read and executed by a computer, the method described in any one of the aforementioned embodiments is executed.

[0048] The beneficial effects of the real-time monitoring system and method for boiler heating surface based on big data described in the present invention are: (1) the system and method predict the temperature of the boiler heating surface through big data, and the prediction accuracy is high. (2) Since the system and method can accurately predict the temperature response of the boiler heating surface under random load fluctuations, it is not only suitable for conventional thermal power units with regular load fluctuations, but also suitable for thermal power units with frequent and large random load fluctuations, which are used for peak regulation of new energy such as solar energy and wind power, and have broad application prospects. (3) The system and method have a small amount of calculation and can realize real-time prediction of the heating surface temperature. (4) The system and method are based on pure software algorithms and can be easily integrated into the existing real-time monitoring system for boiler heating surface temperature. (5) The system and method combine the advantages of big data and expert systems. With the accumulation of operating big data, the accuracy of the expert system output will be further improved, and the prediction effect of the system will be continuously optimized. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 This is a structural diagram of a real-time monitoring system for boiler heating surfaces based on big data disclosed in the present invention;

[0050] Figure 2 The present invention discloses a flow chart of a method for real-time monitoring of boiler heating surfaces based on big data. DETAILED DESCRIPTION

[0051] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not used to limit the scope of the present invention.

[0052] It is understandable that the technical terms, English abbreviations, etc. that appear in the following text are all prior art, and those skilled in the art can understand their meanings based on the context, and will not be explained in detail in this article due to space constraints.

[0053] In the description of the present application, “plurality” means two or more than two, unless otherwise clearly and specifically defined.

[0054] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0055] The present invention provides a real-time monitoring system and method for boiler heating surface based on big data, which are described below respectively.

[0056] A specific embodiment of the present invention discloses a real-time monitoring system for boiler heating surfaces based on big data, such as Figure 1 As shown, including:

[0057] The unit load prediction unit 1 is used to predict the unit load change of the boiler in the future period of time according to the current data of external factors, and output the unit load change curve.

[0058] The fast Fourier transform (FFT) calculation unit 2 is used to transform the unit load variation curve from the time domain to the frequency domain.

[0059] The boiler heating surface temperature response calculation unit 3 is used to calculate the temperature change response of the heating surface under the load changes of the units with different frequencies.

[0060] The inverse fast Fourier transform (IFFT) calculation unit 4 is used to transform the temperature change response of the boiler heating surface from the frequency domain to the time domain, fit it into a continuous curve, and output it.

[0061] Furthermore, in a preferred embodiment, the unit load prediction unit includes a data processing module and a unit load prediction module, wherein the data processing module cleans, screens, removes duplicates, formats and normalizes the input current data of external factors and outputs the data to the unit load prediction module.

[0062] Furthermore, in a preferred embodiment, the unit load forecasting adopts BP neural network to forecast the unit load, and combines LM algorithm to improve the forecasting accuracy and prevent the BP neural network from falling into the local optimum.

[0063] When the BP neural network is combined with big data and applied to unit load forecasting, it has the advantages of good robustness, self-learning, and high prediction accuracy. Compared with the traditional method based on historical load fitting, the BP neural network prediction is significantly more accurate. Since the external data used to predict the unit load in this embodiment has a high dimension and contains highly random external data such as temperature, light, and wind speed, the BP neural network converges slowly during the training process and is prone to fall into the local optimum. This embodiment is combined with the LM algorithm. On the one hand, it can increase the iteration span of the BP network weight coefficient W(k) to prevent the unit load forecast from falling into the local optimum in advance. On the other hand, it can accelerate the convergence of training results, reduce the amount of calculation, and improve the real-time performance of unit load forecasting.

[0064] Furthermore, in a preferred embodiment, the BP neural network is trained through external factors and historical statistical data of unit load.

[0065] Furthermore, in a preferred embodiment, the boiler heating surface temperature response calculation unit comprises an input filtering unit, a temperature response sub-calculation unit, and a response synthesis unit. The input filtering unit filters the load prediction spectrum and inputs it to the temperature response calculation sub-unit, and the temperature response calculation sub-unit inputs the temperature response spectrum to the response synthesis unit after completing the temperature response calculation.

[0066] Specifically, the temperature response sub-calculation unit is divided according to the position of the heating surface, and the temperature response at different positions is calculated respectively. In a preferred embodiment, the boiler heating surface is further divided into different sub-heating surfaces according to the position, such as water-cooled wall, superheater, reheater, economizer, etc. Each sub-heating surface is further divided into different temperature response sub-calculation units according to the number of the pipeline, for example, superheater 1# panel 1# tube temperature response sub-calculation unit, superheater 1# panel 2# tube temperature response sub-calculation unit, etc.

[0067] Furthermore, in a preferred embodiment, the temperature response calculation subunit calculates the temperature response of the heating surface using an expert system supported by big data.

[0068] According to the principle of thermal power units, when the unit load increases, the temperatures of different heating surfaces of the boiler will undergo the following response changes.

[0069] Rapid response stage: When the unit load increases, the unit's control system first increases the amount of boiler fuel and air volume to increase the combustion intensity, but this will cause the furnace temperature to rise in a short period of time, and then be transferred to each heating surface by radiation and convection. In the early stage, due to the enhanced fuel combustion, the temperature of the radiation heating surface (such as the water-cooled wall) will rise rapidly, but the temperature response of the convection heating surface (such as the superheater and reheater) will be relatively lagging, because it takes a certain amount of time for the heat to be transferred to the convection heating surface.

[0070] Transition stage: As the load continues to increase, the temperature and flow rate of the convection flue gas gradually increase, and the convection heating surface begins to show a significant temperature increase.

[0071] Stabilization stage: After adjustment, when the boiler system reaches a new equilibrium state, the temperature of all heating surfaces will stabilize and match the new load level. At this time, the temperature of both the radiation heating surface and the convection heating surface will stabilize at a higher level to adapt to the increased load.

[0072] Therefore, it can be seen that, ignoring the influence of random disturbance factors, when the initial conditions of the boiler are constant, such as the feed water flow, feed water temperature, outlet steam temperature, outlet steam pressure and other parameters are determined, the thermal power unit will produce an approximate periodic response to the boiler heating surface temperature under the excitation of a certain frequency load fluctuation. Therefore, the temperature of the boiler heating surface can be predicted by the unit load spectrum. The statistical data of the existing units also proves this point.

[0073] In addition, the present invention also provides a specific embodiment, disclosing a real-time monitoring method for boiler heating surface based on big data, such as Figure 2 As shown, the method comprises the following steps:

[0074] S-1 collects factors that affect unit load and organizes and cleans the data, and integrates the data into a unified format.

[0075] S-2: Extract data features from the data and identify key features that affect unit load forecasting.

[0076] S-3: The unit load prediction unit 1 outputs the unit load prediction curve according to the key features.

[0077] S-4: The fast Fourier transform (FFT) calculation unit converts the boiler load prediction curve in the time domain into the unit load prediction spectrum in the frequency domain through the discrete Fourier transform algorithm.

[0078] S-5: The boiler heating surface temperature response calculation unit calculates and outputs the periodic response of the boiler heating surface temperature to unit load fluctuations of different frequencies based on the boiler predicted load spectrum, and outputs the boiler heating surface temperature response spectrum.

[0079] S-5: The boiler heating surface temperature response spectrum is converted into a discrete curve of the heating surface temperature response in the time domain through the inverse fast Fourier transform IFFT calculation unit.

[0080] S-6: After the inverse transformation, the time-domain discrete curve of the temperature response of the heated surface is fitted into a continuous curve through the fitting algorithm and output.

[0081] Compared with the existing technology, the advantages of the method are: introducing a big data method, selecting key external factors that affect the unit load, effectively improving the prediction accuracy of the unit load, and thus improving the accuracy of the calculation of the heating surface temperature response; calculating the boiler heating surface temperature response through the unit load prediction curve, and realizing real-time prediction of the heating surface temperature; with the help of fast Fourier transform, the boiler heating surface temperature can still be accurately predicted when the load fluctuates randomly and significantly.

[0082] Furthermore, in a preferred real-time example, in the S-1 step, multiple external factors that affect the unit load are first preliminarily selected, and short-term factors include time, temperature, electricity price, light, wind, humidity, rainfall, snowfall, holidays, etc.; long-term factors include seasonal factors, regional population, urbanization rate, GDP, primary / secondary and tertiary industry proportions, etc., and the data is cleaned, screened, deduplicated and formatted. Specifically, in the formatting step, the response variable, i.e., the historical statistical data of the unit load, yi, is normalized, and the historical statistical data of the external factors, x, is normalized. ij After standardization, the data y i and x ij Satisfy the following formula.

[0083]

[0084] Furthermore, in a preferred embodiment, the S-2 step analyzes the relationship between external factors and unit load through the LASSO algorithm, removes factors with poor correlation, and finally selects six variables including time, electricity price, temperature, seasonal factors, population, and GDP as external data input into the unit load forecasting unit.

[0085] Furthermore, the transformation formula adopted by the S-4 fast Fourier transform FFT calculation unit is:

[0086]

[0087] In the formula, X(K) is the unit load in the frequency domain, and the frequency is Taking into account the spectrum accuracy of the unit, preferably, the sampled signal length is selected to be 400 minutes, and the spectrum accuracy of the unit is f=1 / 24000=416 μHz.

[0088] Furthermore, in a preferred embodiment, the step S-5 can be further divided into the following steps:

[0089] Step 1: The filtering unit is used to filter the input unit load prediction curve.

[0090] Step 2: The temperature response calculation subunit calculates the temperature response.

[0091] Step 3: Synthesis of the temperature response spectrum of the heated surface.

[0092] According to the actual operation data, the unit load prediction spectrum is concentrated near the low-frequency fundamental wave. As the frequency increases, the amplitude decreases rapidly. When the frequency is greater than 3 times the fundamental wave frequency, the amplitude of the unit load is less than 10% of the fundamental wave amplitude. In addition, under high-frequency load fluctuations, the unit combustion is unstable, and the prediction deviation according to the excitation response model is large. Therefore, in a preferred embodiment, the filtering unit in step 1 filters the unit load prediction spectrum to eliminate the load prediction spectrum above 3 times the frequency.

[0093] Furthermore, in a preferred embodiment, the step 2 can be further divided into the following steps:

[0094] Establish an expert database based on the unit load test results.

[0095] The unit load frequency and amplitude, as well as the initial conditions of the boiler are input into each temperature response subunit, which calculates the response spectrum of the frequency load fluctuation at each position of the heating surface based on the expert system.

[0096] outputting the temperature response spectrum to a response synthesis unit;

[0097] Further, in a preferred embodiment, the method for establishing an expert database according to the unit load test results is as follows:

[0098] Through the test, the response of the pipes at different positions of the boiler heating surface to load fluctuations is recorded. Specifically, before the boiler is put into operation, a unit test platform with electrical load is established, a set of initial parameters for the unit operation is selected, and the output power of the unit is adjusted through the electrical load to make the boiler load change according to the specified characteristic frequency and amplitude. After the system reaches a steady state, the temperature of each pipeline of the heating surface is measured by the sensor installed on the heating surface and the temperature response cycle curve is recorded. The recorded heating surface temperature response change curve is subjected to spectrum analysis to obtain the temperature response spectrum of the heating surface pipeline.

[0099] The present invention also provides a specific embodiment, further explaining how to obtain the temperature response spectrum at each position of the heating surface in the expert database. Take the measurement of the temperature response of the superheater 3# screen 5# tube to the load fluctuation of 4160μHz as an example. First, the unit is adjusted initially, and the inlet water temperature is 291℃, the superheated steam temperature is 571℃, the superheated steam pressure is 25.1Mpa, the superheated steam flow is 1150t / h, and the output end of the boiler generator is connected to the simulated RL load, and the power factor is 0.9. Then, the power is kept unchanged and the RL load is adjusted so that the output active power of the boiler fluctuates with a sine wave of 4160μHz, and the amplitude of the load fluctuation is 25MW. At the same time, the temperature change of the superheater 3# screen 5# tube is measured and recorded by the thermocouple installed on the heating surface pipeline. After the recording is completed, the temperature recording result is input into the spectrum analyzer for analysis, and the record includes the fundamental wave and the harmonics with an amplitude greater than 20% of the fundamental wave. Then the inlet water temperature, superheated steam pressure, superheated steam temperature, superheated steam pressure, load fluctuation amplitude, fundamental frequency, and harmonic frequency form a sample X 11 , recorded in the expert database.

[0100] Then continue to change the fluctuation amplitude of the unit, for example, reduce the fluctuation amplitude to 34MW, repeat the above steps, and record a sample data X again. 12 Repeat the above operation i times to obtain the sample matrix [X 11 ,X 12 ,...X 1i ].

[0101] After recording all the load change sample data, change the load fluctuation frequency to 4560μHz, and record a set of sample matrices [X 21 ,X 22 ,...X 2i ]

[0102] Then continue to change the initial conditions to other working conditions, repeat the above steps and record a new set of sample data.

[0103] After the above test process, the expert database has accumulated sample data of the temperature response of the heating surface under all different working conditions, load fluctuation frequencies and amplitudes.

[0104] Furthermore, a method for calculating the temperature response by the temperature response calculation subunit is described in combination with an actual example. The subunit calculates the temperature response of the superheater 3# screen 5# tube in the following manner:

[0105] Input the initial conditions of the boiler, the inlet water temperature is 291℃, the superheated steam temperature is 571℃, the superheated steam pressure is 25.1Mpa, the current state of the boiler, and then according to the unit load fluctuation spectrum, input a load fluctuation with a frequency of 4160μHz and an amplitude of 15MW. The expert system outputs the temperature response spectrum T1(ω) of the heating surface 3# screen 5# tube according to the corresponding expert database. Then input the load fluctuation with a frequency of 7320μHz and an amplitude of 8MW, and the expert system outputs the temperature response spectrum T2(ω) of the heating surface 3# screen 5# tube according to the corresponding expert database.

[0106] Further, in a preferred embodiment, a spectrum synthesis method in step three is:

[0107] According to the approximate model, the response synthesis unit calculates the temperature response T under different frequency load fluctuation excitation according to the following formula: i (ω i ) for superposition synthesis.

[0108] T=k1T1(ω1)+k2T2(ω2)+k3T3(ω3)+...+k n T n (ω n )

[0109] where k i is the fitting superposition correction coefficient of the temperature response spectrum corresponding to the i-th load fluctuation, and T is the final synthetic spectrum of the pipeline. The response synthesis unit can use a variety of methods to obtain the fitting superposition coefficient. Preferably, the fitting superposition correction is obtained by fitting using the least squares method, and the steps of obtaining are as follows:

[0110] First, according to the measured spectrum of the temperature response of the pipeline, T=[T1,T2,...T i ], construct a sample with a total number of samples of m, and define the error function:

[0111]

[0112] Let S′(k) = 0 and solve k by gradient descent method. j (j=1...n), thereby obtaining the fitting superposition correction coefficient of the response synthesis unit.

[0113] Compared with the prior art, the present invention uses fast Fourier transform (FFT) to decompose the irregular unit load prediction curve into sinusoidal excitations of different frequencies in the frequency domain, and calculates the boiler heating surface temperature responses corresponding to the unit loads of different frequencies through the boiler heating surface temperature response unit and performs response synthesis to obtain a complete response spectrum of the heating surface temperature. Then, the heating surface temperature response curve in the time domain is obtained through inverse fast Fourier transform, thereby realizing the real-time prediction function of the heating surface temperature.

[0114] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by a technician familiar with the technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. A real-time monitoring system for boiler heating surfaces based on big data, characterized in that: The system includes a unit load prediction unit, a fast Fourier transform FFT calculation unit, a boiler heating surface temperature response calculation unit, and an inverse fast Fourier transform IFFT calculation unit; The unit load prediction unit is used to predict the unit load change of the boiler in the future period of time according to the current data of external factors, and output the unit load change curve; The fast Fourier transform (FFT) calculation unit is used to transform the unit load change curve from the time domain to the frequency domain; The boiler heating surface temperature response calculation unit is used to calculate the temperature response of the heating surface under load fluctuations of different frequencies; The inverse fast Fourier transform (IFFT) calculation unit is used to transform the temperature response of the boiler heating surface from the frequency domain to the time domain and fit it into a continuous curve output.

2. The real-time monitoring system for boiler heating surface based on big data according to claim 1 is characterized in that: The unit load prediction unit includes a data processing module and a unit load prediction module; wherein the data processing module cleans, screens, removes duplicates, formats and normalizes the input current data of external factors, and then outputs it to the unit load prediction module.

3. The real-time monitoring system for boiler heating surface based on big data according to claim 1 is characterized in that: The boiler heating surface temperature response calculation unit includes an input filtering unit, a temperature response calculation subunit, and a response synthesis unit; wherein the input filtering unit filters the load prediction spectrum and inputs it into the temperature response calculation subunit, and the temperature response calculation subunit inputs the temperature response spectrum into the response synthesis unit after completing the temperature response calculation.

4. The real-time monitoring system for boiler heating surface based on big data according to claim 2 is characterized in that: The unit load prediction module adopts BP neural network to predict unit load, and combines LM algorithm to improve prediction accuracy and prevent BP neural network from falling into local optimal point.

5. The real-time monitoring system for boiler heating surface based on big data according to claim 4 is characterized in that: The BP neural network is trained through external factors and historical statistical data of unit load.

6. The real-time monitoring system for boiler heating surface based on big data according to claim 3 is characterized in that: The temperature response calculation subunit calculates the temperature response using an expert system supported by big data.

7. A real-time monitoring method for boiler heating surface based on big data, characterized in that: The method predicts the temperature of the heating surface by the following steps: Step 1: Analyze the factors affecting the unit load and collect, clean and complete the data to unify the data format; Step 2: Extract data features from the data and identify key features that affect unit load forecasting; Step 3: Unit load prediction Output unit load prediction curve according to key features; Step 4: The fast Fourier transform (FFT) calculation unit converts the boiler load prediction curve in the time domain into the unit load prediction spectrum in the frequency domain through the discrete Fourier transform algorithm; Step 5: The boiler heating surface temperature response calculation unit calculates and outputs the periodic response of the boiler heating surface temperature to unit load fluctuations of different frequencies according to the boiler load prediction spectrum, and outputs the boiler heating surface temperature response spectrum; Step 6: The boiler heating surface temperature response spectrum is inversely transformed into a heating surface temperature response discrete curve in the time domain through an inverse fast Fourier transform IFFT calculation unit; Step 7: Use the fitting algorithm to fit the discrete curve of the heated surface temperature response into a continuous curve output.

8. The method for real-time monitoring of boiler heating surfaces based on big data according to claim 7 is characterized in that: The second step is to analyze the relationship between external factors and unit load by using LASSO algorithm, and remove factors with poor correlation.

9. The method for real-time monitoring of boiler heating surfaces based on big data according to claim 7 is characterized in that: The step five comprises the following steps: Step 5.1: the filtering unit filters the input unit load prediction curve; Step 5.2: the temperature response calculation subunit calculates the temperature response; Step 5.3 Synthesis of the temperature response spectrum of the heated surface.

10. The method for real-time monitoring of boiler heating surfaces based on big data according to claim 9, characterized in that: The temperature response calculation subunit in step 5.2 calculates the temperature response, including the following steps: Establish an expert database based on the unit load test results; The unit load frequency and amplitude, as well as the initial conditions of the boiler are input into each temperature response subunit, and the subunit calculates the response spectrum of the frequency load fluctuation at each position of the heating surface based on the expert system; The temperature response spectrum is output to the response synthesis unit.

Images