Intelligent recovery method and system for chemical production waste gas based on digital twinning

Abstract

The application discloses a chemical production waste gas intelligent recovery method and system based on digital twinning, and relates to the related field of gas separation and purification, wherein the method comprises the following steps: acquiring a plurality of key devices and laying out multidimensional sensors; traversing a plurality of sensor arrays to collect multi-source waste gas process data and perform uncertainty preprocessing within a preset monitoring window; calling a chemical production waste gas treatment digital twinning model, combining a plurality of centralized monitoring data arrays and a plurality of data uncertainty coefficient arrays to perform double uncertainty integration from two dimensions of semantic uncertainty and working condition uncertainty; and adaptively adjusting real-time waste gas recovery process parameters according to integrated chemical production waste gas recovery working condition characteristics. The application solves the technical problems of insufficient process adjustment precision, fluctuating recovery efficiency and unstable emission quality of existing waste gas recovery, and achieves the technical effects of improving process adjustment precision, improving recovery efficiency and guaranteeing emission quality.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of gas separation and purification, in particular to a chemical production waste gas intelligent recovery method and system based on digital twinning. BACKGROUND

[0002] Efficient recovery and standard emission of halogenated aromatic chemical production waste gas are of great significance to ecological environment protection and resource recycling. At present, the industry mainly adopts the combined process of acid gas washing, low-temperature condensation and activated carbon adsorption regeneration, and processes waste gas through preset fixed process parameters or manual experience adjustment. This method cannot capture dynamic changes such as waste gas composition fluctuations and equipment operation minor abnormalities in real time, and does not consider the uncertainty in data collection and working condition prediction, resulting in unstable recovery efficiency and emission quality easy to exceed standard.

[0003] At present, the related technology in the art has the technical problem of insufficient process adjustment precision of chemical production waste gas recovery, resulting in fluctuation of recovery efficiency and instability of emission quality. SUMMARY

[0004] The present application provides a chemical production waste gas intelligent recovery method and system based on digital twinning, which adopts a chemical waste gas treatment system including acid gas washing, multi-stage condensation and activated carbon adsorption regeneration units, arranges multi-dimensional sensors at key equipment to construct sensor arrays, traverses each array to collect multi-source waste gas process data in a preset monitoring window and complete uncertainty preprocessing, obtains a centralized monitoring data array and a data uncertainty coefficient array, calls a pre-constructed chemical waste gas treatment digital twinning model, integrates the working condition characteristics from the semantic and working condition dual uncertainty dimensions combined with the two types of arrays, adjusts the real-time process parameters according to the characteristics, applies the adjusted parameters to the actual system to complete process parameter adjustment, and solves the technical problem of insufficient process adjustment precision of existing chemical production waste gas recovery, resulting in fluctuation of recovery efficiency and instability of emission quality, and achieves the technical effects of improving process adjustment precision, improving recovery efficiency and ensuring emission quality.

[0005] The application provides a chemical production waste gas intelligent recovery method based on digital twinning, comprising: acquiring a plurality of key devices of a chemical production waste gas treatment system, arranging a multi-dimensional sensor at the plurality of key devices, and obtaining a plurality of sensor arrays, wherein the chemical production waste gas treatment system comprises an acid gas washing unit, a multi-stage condensing unit, and an activated carbon adsorption regeneration unit; traversing the plurality of sensor arrays to perform multi-source waste gas process data acquisition and uncertainty preprocessing within a preset monitoring window to obtain a plurality of centralized monitoring data arrays and a plurality of data uncertainty coefficient arrays; calling a pre-constructed digital twinning model of the chemical production waste gas treatment, combining the plurality of centralized monitoring data arrays and the plurality of data uncertainty coefficient arrays to perform double uncertainty integration from two dimensions of semantic uncertainty and working condition uncertainty to obtain integrated chemical production waste gas recovery working condition characteristics; adaptively adjusting real-time waste gas recovery process parameters according to the integrated chemical production waste gas recovery working condition characteristics to obtain adjusted waste gas recovery process parameters, and adjusting the recovery process parameters of the chemical production waste gas treatment system based on the adjusted waste gas recovery process parameters.

[0006] In a possible implementation, the plurality of sensor arrays are traversed to perform multi-source waste gas process data acquisition and uncertainty preprocessing within a preset monitoring window to obtain a plurality of centralized monitoring data arrays and a plurality of data uncertainty coefficient arrays, and the following processing is performed: the plurality of sensor arrays are traversed to perform multi-source waste gas process data acquisition within a preset monitoring window to obtain a plurality of monitoring data sequence arrays; a first monitoring data sequence in the plurality of monitoring data sequence arrays is extracted, and the first monitoring data sequence is subjected to uncertainty preprocessing to obtain first centralized monitoring data and a first data uncertainty coefficient, wherein the uncertainty preprocessing comprises data centralization screening and uncertainty analysis; and the first centralized monitoring data and the first data uncertainty coefficient are added to the plurality of centralized monitoring data arrays and the plurality of data uncertainty coefficient arrays.

[0007] In a possible implementation, the first monitoring data sequence in the plurality of monitoring data sequence arrays is extracted, and the first monitoring data sequence is subjected to uncertainty preprocessing to obtain first centralized monitoring data and a first data uncertainty coefficient, wherein the uncertainty preprocessing comprises data centralization screening and uncertainty analysis, and the following processing is performed: a data mean value of the first monitoring data sequence is calculated to obtain an initial data centralization screening center; the first monitoring data sequence is subjected to mean shift screening with the initial data centralization screening center as a starting point to determine the first centralized monitoring data; and the first monitoring data sequence is subjected to uncertainty analysis based on the first centralized monitoring data to determine the first data uncertainty coefficient.

[0008] In a possible implementation, based on the first centralized monitoring data, uncertainty analysis is performed on the first monitoring data sequence to determine a first data uncertainty coefficient, and the following processing is performed: data screening is performed on the first monitoring data sequence according to a preset similarity bandwidth, and an initial deterministic neighborhood of the first centralized monitoring data is constructed, where the initial deterministic neighborhood includes monitoring data in the first monitoring data sequence that has a similarity to the first centralized monitoring data within a preset similarity bandwidth range; an edge of the initial deterministic neighborhood is diffused according to 1 / 2 of the preset similarity bandwidth to obtain a diffused deterministic neighborhood; it is judged whether a neighborhood data quantity difference between the diffused deterministic neighborhood and the initial deterministic neighborhood is greater than or equal to a preset data quantity difference threshold, and if yes, the initial deterministic neighborhood is taken as a target deterministic neighborhood; a data quantity difference between the target deterministic neighborhood and the first monitoring data sequence is calculated, and a difference between a calculation result and 1 is taken as the first data uncertainty coefficient.

[0009] In a possible implementation, the following processing is performed: if the neighborhood data quantity difference between the diffused deterministic neighborhood and the initial deterministic neighborhood is less than the preset data quantity difference threshold, the edge of the diffused deterministic neighborhood is diffused according to 1 / 2 of the preset similarity bandwidth to obtain an iterative diffused deterministic neighborhood, and the processing is iterated until a neighborhood data quantity difference between two adjacent diffusion times is greater than or equal to the preset data quantity difference threshold, and a target deterministic neighborhood is obtained.

[0010] In a possible implementation, the following processing is performed: the process flow chain and the key equipment structure of the acid gas washing unit, the multi-stage condensation unit and the activated carbon adsorption regeneration unit in the chemical production waste gas treatment system are obtained, a digital twin simulator is called for twin simulation to obtain a chemical production waste gas treatment digital twin model, and the chemical production waste gas treatment digital twin model includes: a two-stage acid gas washing tower mass transfer model for simulating an acid gas removal process; a multi-stage low-temperature condensation model for simulating a halogenated aromatic compound condensation capture process; an activated carbon adsorption-desorption bed axial model, a mass transfer model and a heat release model.

[0011] In a possible implementation, a pre-constructed chemical production waste gas treatment digital twin model is called, combined with the plurality of centralized monitoring data arrays and the plurality of data uncertainty coefficient arrays, double uncertainty integration is performed from two dimensions of semantic uncertainty and working condition uncertainty, integrated chemical production waste gas recovery working condition characteristics are obtained, and the following processing is performed: the plurality of centralized monitoring data arrays are synchronously transmitted to the chemical production waste gas treatment digital twin model for twin simulation prediction, digital twin simulation prediction working condition characteristics are obtained; semantic uncertainty and working condition uncertainty analysis are performed in combination with the plurality of data uncertainty coefficient arrays and the digital twin simulation prediction working condition characteristics, semantic uncertainty coefficients and working condition uncertainty coefficients are obtained; the integrated uncertainty coefficient is obtained based on the semantic uncertainty coefficients and the working condition uncertainty coefficients, and the integrated uncertainty coefficient is used to identify the digital twin simulation prediction working condition characteristics, and the integrated chemical production waste gas recovery working condition characteristics are obtained.

[0012] In a possible implementation, the following processing is performed: the semantic uncertainty coefficients are obtained by combining the plurality of data uncertainty coefficient arrays and performing weighting; the digital twin simulation prediction working condition characteristics are matched with a historical abnormal chemical production waste gas recovery scene library, a plurality of matching historical abnormal chemical production waste gas recovery scenes with a matching similarity greater than or equal to a preset matching similarity threshold are extracted; and the ratio of the number of the plurality of matching historical abnormal chemical production waste gas recovery scenes to the total number of scenes in the historical abnormal chemical production waste gas recovery scene library is obtained, and the working condition uncertainty coefficient is obtained.

[0013] In a possible implementation, the following processing is performed: the integrated uncertainty coefficient of the integrated chemical production waste gas recovery working condition characteristics is extracted, the integrated uncertainty coefficient is compared with a standard uncertainty coefficient, and a single adjustment bandwidth is determined; the real-time waste gas recovery process parameters are adaptively adjusted based on the single adjustment bandwidth, and stage-adjusted waste gas recovery process parameters are obtained; the stage-adjusted waste gas recovery process parameters are adaptively simulated and verified by using the chemical production waste gas treatment digital twin model, and if the verification is passed, the stage-adjusted waste gas recovery process parameters are used as adjusted waste gas recovery process parameters.

[0014] The application also provides a chemical production waste gas intelligent recovery system based on digital twinning, comprising: a sensor arrangement module for obtaining a plurality of key devices of a chemical production waste gas treatment system, arranging multi-dimensional sensors at the plurality of key devices, and obtaining a plurality of sensor arrays, wherein the chemical production waste gas treatment system comprises an acid gas washing unit, a multi-stage condensing unit, and an activated carbon adsorption regeneration unit; a data acquisition module for traversing the plurality of sensor arrays to perform multi-source waste gas process data acquisition and uncertainty preprocessing within a preset monitoring window, obtaining a plurality of centralized monitoring data arrays and a plurality of data uncertainty coefficient arrays; a dual uncertainty integration module for calling a pre-constructed digital twinning model of the chemical production waste gas treatment system, combining the plurality of centralized monitoring data arrays and the plurality of data uncertainty coefficient arrays to perform dual uncertainty integration from two dimensions of semantic uncertainty and working condition uncertainty, and obtaining integrated chemical production waste gas recovery working condition characteristics; a recovery process parameter adjustment module for adaptively adjusting real-time waste gas recovery process parameters according to the integrated chemical production waste gas recovery working condition characteristics, obtaining adjusted waste gas recovery process parameters, and adjusting the recovery process parameters of the chemical production waste gas treatment system based on the adjusted waste gas recovery process parameters.

[0015] The chemical production waste gas intelligent recovery method and system based on digital twinning provided by the application first obtain a plurality of key devices of a chemical production waste gas treatment system, arrange multi-dimensional sensors at the plurality of key devices, and obtain a plurality of sensor arrays, wherein the chemical production waste gas treatment system comprises an acid gas washing unit, a multi-stage condensing unit, and an activated carbon adsorption regeneration unit. Then, the plurality of sensor arrays are traversed to perform multi-source waste gas process data acquisition and uncertainty preprocessing within a preset monitoring window, a plurality of centralized monitoring data arrays and a plurality of data uncertainty coefficient arrays are obtained. Next, a pre-constructed digital twinning model of the chemical production waste gas treatment system is called, the plurality of centralized monitoring data arrays and the plurality of data uncertainty coefficient arrays are combined to perform dual uncertainty integration from two dimensions of semantic uncertainty and working condition uncertainty, and integrated chemical production waste gas recovery working condition characteristics are obtained. Finally, real-time waste gas recovery process parameters are adaptively adjusted according to the integrated chemical production waste gas recovery working condition characteristics, adjusted waste gas recovery process parameters are obtained, and the recovery process parameters of the chemical production waste gas treatment system are adjusted based on the adjusted waste gas recovery process parameters. Through the above process, the method and system provided by the application achieve the technical effects of improving process adjustment accuracy, improving recovery efficiency, and ensuring emission quality. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings of the embodiments of the present application will be briefly introduced below, and the flowcharts are used to illustrate the operations performed by the system according to the embodiments of the present application in the present application. It should be understood that the foregoing or the following operations are not necessarily performed in sequence. On the contrary, various steps can be processed in reverse order or simultaneously according to needs. At the same time, other operations can be added to these processes, or one or more steps of operations can be removed from these processes.

[0017] Figure 1 A flowchart of a chemical production waste gas intelligent recycling method based on digital twinning provided by the embodiments of the present application is shown.

[0018] Figure 2 A structural diagram of a chemical production waste gas intelligent recycling system based on digital twinning provided by the embodiments of the present application is shown.

[0019] Legend: sensor arrangement module 10, data acquisition module 20, double uncertainty integration module 30, recycling process parameter adjustment module 40. DETAILED DESCRIPTION

[0020] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined object of the present application, the specific embodiments, structures, features and effects according to the present application will be described in detail below in combination with the drawings and preferred embodiments.

[0021] The embodiments of the present application provide a chemical production waste gas intelligent recycling method based on digital twinning, as shown in Figure 1 The method comprises the following steps:

[0022] In step S100, a plurality of key devices of a chemical production waste gas treatment system are acquired, a plurality of multi-dimensional sensors are arranged at the plurality of key devices, and a plurality of sensor arrays are obtained, wherein the chemical production waste gas treatment system comprises an acid gas washing unit, a multi-stage condensing unit and an activated carbon adsorption regeneration unit.

[0023] Specifically, a data-aware network of the exhaust gas treatment system is built, and monitoring points are determined for key equipment in the three core functional units of the system. For example, the key equipment of the acid gas washing unit includes the washing tower, the circulating pump, and the pH adjusting tank, the multi-stage condensing unit includes the first-stage condenser, the second-stage condenser, and the gas-liquid separator, and the activated carbon adsorption and regeneration unit includes the adsorption tower, the desorption heater, and the vacuum pump. Multidimensional sensors cover key parameters such as temperature, pressure, flow rate, concentration, pH value, and liquid level. For example, a pH sensor and an HC1 / HBr concentration sensor are arranged at the outlet of the washing tower, temperature and pressure sensors are arranged at the inlet and outlet of the condenser, and a halogenated aromatic hydrocarbon VOCs concentration sensor is arranged at the inlet and outlet of the adsorption tower. A combination of multiple types of sensors at each key equipment forms a sensor array, such as a washing tower sensor array that includes a pH sensor, a liquid level sensor, a circulating liquid flow rate sensor, and an outlet acid gas concentration sensor. The sensor arrays of all equipment collectively constitute the data acquisition network of the entire system.

[0024] In step S200, multi-source exhaust gas process data acquisition and uncertainty preprocessing are performed on the multiple sensor arrays within a preset monitoring window to obtain multiple centralized monitoring data arrays and multiple data uncertainty coefficient arrays.

[0025] Specifically, all sensor arrays are traversed, and data is continuously acquired according to a preset monitoring window, such as a fixed time interval of 5 minutes / once or 10 minutes / once, or a fixed processing volume interval of 100 cubic meters of exhaust gas / once, to form an original data set containing multiple types of parameters such as temperature, pressure, and concentration. Uncertainty preprocessing is divided into two parts: data set screening and uncertainty analysis. The data set screening removes abnormal values in the original data, such as jump values caused by sensor failure and transient interference values, and retains core data with statistical representation. The uncertainty analysis analyzes the fluctuation range and reliability of the screened data, and calculates the uncertainty coefficient of the data itself. Finally, each sensor array outputs a set of filtered centralized monitoring data, such as the average value, maximum value, and minimum value of the VOCs concentration of an adsorption tower within 1 hour, and a set of uncertainty coefficients reflecting the reliability of the data. The results of all sensor arrays are respectively aggregated into a centralized monitoring data array and a data uncertainty coefficient array.

[0026] In a possible implementation, the plurality of sensor arrays are traversed to collect multi-source waste gas process data and perform uncertainty preprocessing in a preset monitoring window, to obtain a plurality of centralized monitoring data arrays and a plurality of data uncertainty coefficient arrays, and step S200 further includes step S210 of collecting multi-source waste gas process data in the preset monitoring window by traversing the plurality of sensor arrays, to obtain a plurality of monitoring data sequence arrays. Specifically, all sensors in each sensor array are sequentially subjected to data collection according to the preset monitoring window. For example, for the scrubber sensor array, three types of parameters, pH value, circulating liquid flow, and outlet HCl concentration, are collected, and one value is recorded every 2 minutes, and 3 groups of 15 data points form a sequence after 30 minutes. The multi-type parameter data sequences collected by each sensor array collectively constitute a monitoring data sequence array, and the monitoring data sequence arrays of all sensor arrays are summarized to form a plurality of monitoring data sequence arrays.

[0027] Step S220 extracts a first monitoring data sequence from the plurality of monitoring data sequence arrays, and performs uncertainty preprocessing on the first monitoring data sequence to obtain first centralized monitoring data and a first data uncertainty coefficient, wherein the uncertainty preprocessing includes data centralization screening and uncertainty analysis. Specifically, any one data sequence is selected from the plurality of monitoring data sequence arrays as the first monitoring data sequence, for example, the outlet VOCs concentration data sequence of the adsorption tower. The data centralization screening adopts a statistical method to eliminate outliers, for example, by the 3σ criterion, the average value and the standard deviation of the sequence are calculated, and the data outside the range of ±3 times the standard deviation of the average value is eliminated. The uncertainty analysis calculates the dispersion degree of the screened data, for example, the coefficient of variation of the data is calculated, and the coefficient of variation is converted into the first data uncertainty coefficient. Finally, the screened data is calculated by a statistical method, such as taking the average value, the median or the weighted average value, to obtain the first centralized monitoring data, and the first data uncertainty coefficient is determined.

[0028] Step S230 adds the first centralized monitoring data and the first data uncertainty coefficient into the plurality of centralized monitoring data arrays and the plurality of data uncertainty coefficient arrays. Specifically, a dynamic storage mode of array or matrix is adopted, and a two-dimensional storage structure is preset for the plurality of centralized monitoring data arrays and the plurality of data uncertainty coefficient arrays, respectively, with the row corresponding to the sensor array number and the column corresponding to the data sequence type.

[0029] For example, the first monitoring data sequence is the HCl concentration sequence at the outlet of a washing tower. After processing by S220, the first centralized monitoring data and the first data uncertainty coefficient are obtained. At this time, according to the sensor array number 001 and the data type number C001 to which the data sequence belongs, the first centralized monitoring data is written in the (001, C001) position of the centralized monitoring data array, and the first data uncertainty coefficient is written in the (001, C001) position of the data uncertainty coefficient array. If a second monitoring data sequence of the same sensor array is processed subsequently, the centralized monitoring data and the uncertainty coefficient corresponding to the same sensor array number 001 are written in the data type number position, respectively. The result storage of all monitoring data sequences is completed in turn, and finally the global centralized monitoring data array and the data uncertainty coefficient array covering all sensor arrays and all parameter types are formed.

[0030] In a possible implementation, a first monitoring data sequence in the plurality of monitoring data sequence arrays is extracted, and uncertainty preprocessing is performed on the first monitoring data sequence to obtain first centralized monitoring data and a first data uncertainty coefficient. The uncertainty preprocessing includes data centralization screening and uncertainty analysis. Step S220 further includes step S221 of calculating a data mean value of the first monitoring data sequence to obtain an initial data centralization screening center. Specifically, the arithmetic mean method is used to calculate the average value of all original data in the first monitoring data sequence, and the average value is taken as the screening center. For example, the first monitoring data sequence is the outlet temperature data of a certain condenser: 10℃, 10.2℃, 9.8℃, 10.1℃, 9.9℃, 25℃ (an abnormal value), and 10℃. First, the arithmetic mean value of all 7 data is calculated: (10+10.2+9.8+10.1+9.9+25+10) ÷ 7 ≈ 12.14℃, which is the initial data centralization screening center. Subsequent screening is carried out around the center to determine whether the data is within a reasonable range.

[0031] Step S222, the mean shift filtering is performed on the first monitoring data sequence with the initial filtering center as the starting point to determine the first centralized monitoring data. Specifically, the core of the mean shift filtering is to constantly update the filtering center, gradually eliminate outliers, and finally retain normal data. For example, the initial filtering center is 12.14℃, and the abnormal value is 25℃. First, the filtering range is set, for example, the initial center ± 5℃, at this time the data in the range is 10℃, 10.2℃, 9.8℃, 10.1℃, 9.9℃, 10℃, and the 25℃ is eliminated. Then, the new mean of the 6 data in the range is calculated: (10+10.2+9.8+10.1+9.9+10) ÷ 6=10℃, and the new mean is taken as the new filtering center. The filtering range is set again, for example, the new center ± 0.5℃, and it is judged whether the remaining data is in the range of 9.5℃-10.5℃, and all the data meet the requirements, and the filtering is stopped. Finally, the average of the 6 data 10℃ is taken as the first centralized monitoring data.

[0032] Step S223, the uncertainty of the first monitoring data sequence is analyzed based on the first centralized monitoring data to determine the first data uncertainty coefficient. Specifically, the dispersion index of the normal data after filtering in the first monitoring data sequence is calculated based on the first centralized monitoring data, for example, the dispersion index of 10℃, 10.2℃, 9.8℃, 10.1℃, 9.9℃, 10℃ is calculated based on 10℃. The dispersion index includes standard deviation, variance or coefficient of variation, etc. For example, the standard deviation is calculated: first, the square of the difference between each data and the mean value is calculated (0, 0.04, 0.04, 0.01, 0.01, 0), and then the average value 0.1 ÷ 6≈0.0167 is calculated, and then the square root is taken to get the standard deviation≈0.129. Then the standard deviation is converted into the data uncertainty coefficient, for example, through the formula: uncertainty coefficient=standard deviation ÷ mean, that is, 0.129 ÷ 10≈0.0129, which is the first data uncertainty coefficient. The smaller the coefficient, the more stable the data and the lower the uncertainty.

[0033] In one possible implementation, the uncertainty of the first monitoring data sequence is analyzed based on the first centralized monitoring data to determine the first data uncertainty coefficient, and step S223 further includes step S2231 of performing data filtering on the first monitoring data sequence according to a preset similarity bandwidth to construct an initial deterministic neighborhood of the first centralized monitoring data, wherein the initial deterministic neighborhood includes monitoring data in the first monitoring data sequence having a similarity to the first centralized monitoring data within a preset similarity bandwidth. Specifically, the preset similarity bandwidth is a threshold value set in advance to judge whether the data is similar to the centralized monitoring data, which is expressed in absolute deviation or relative deviation. For example, for the temperature parameter, the preset similarity bandwidth can be an absolute deviation of ± 0.5℃, or for the concentration parameter, the preset similarity bandwidth can be a relative deviation of ± 5%.

[0034] Taking the first centralized monitoring data 10℃ as an example, the data in the first monitoring data sequence that deviates from 10℃ by ±0.5℃, i.e. the data between 9.5℃ and 10.5℃, is screened, and the data that meets the condition is 10℃, 10.2℃, 9.8℃, 10.1℃, 9.9℃ and 10℃. These data collectively constitute an initial deterministic neighborhood, and the data in the neighborhood is considered to be highly similar to the centralized monitoring data and to have high reliability.

[0035] In step S2232, the edges of the initial deterministic neighborhood are diffused by 1 / 2 of the preset similarity bandwidth to obtain a diffused deterministic neighborhood. Specifically, if the preset similarity bandwidth is ±0.5℃, 1 / 2 of the preset similarity bandwidth is ±0.25℃, and the temperature range of the initial deterministic neighborhood is 9.5℃-10.5℃, the upper and lower edges of the range are diffused outward by 0.25℃, i.e. the lower limit is 9.5℃-0.25℃=9.25℃, and the upper limit is 10.5℃+0.25℃=10.75℃, and the new temperature range 9.25℃-10.75℃ is the diffused deterministic neighborhood. At this time, the first monitoring data sequence is checked again, and if there is no data in the sequence that is outside the initial neighborhood but within the diffused neighborhood, such as 9.3℃ and 10.6℃, the data in the diffused neighborhood is the same as that in the initial neighborhood; if there is such data, it is included in the diffused deterministic neighborhood.

[0036] Step S2233, determine whether the difference between the data quantity of the diffusion determined neighborhood and the initial determined neighborhood is greater than or equal to a preset data quantity difference threshold value, if yes, take the initial determined neighborhood as the target determined neighborhood, if the difference between the data quantity of the diffusion determined neighborhood and the initial determined neighborhood is less than the preset data quantity difference threshold value, diffuse the edge of the diffusion determined neighborhood according to 1 / 2 of the preset similarity bandwidth, obtain an iterative diffusion determined neighborhood, and so on until the difference between the data quantity of the adjacent two diffusion neighborhoods is greater than or equal to the preset data quantity difference threshold value, and obtain the target determined neighborhood. Specifically, the preset data quantity difference threshold value is a data quantity standard for judging whether the diffusion is effective, for example, it is set to 1, that is, the data quantity after diffusion is increased by 1 or more than before, which is considered effective. Taking the initial determined neighborhood data quantity of 6 and the diffusion determined neighborhood data quantity of 6 as an example, the difference between the two is 0, which is less than the threshold value 1, and the diffusion needs to be continued: diffuse the edge of the diffusion determined neighborhood 9.25-10.75℃ by ±0.25℃ again, obtain the iterative diffusion determined neighborhood 9-11℃, and check the data sequence again, if there is still no new data, the data quantity difference is still 0, continue to diffuse, until after a certain diffusion, new data is added in the neighborhood, for example, there is 8.9℃ data in the sequence, which is included in the data after diffusion to 8.75-11.25℃, the data quantity changes from 6 to 7, the difference is 1≥threshold value 1, stop iteration, and take the neighborhood 9-11℃ after the previous diffusion as the target determined neighborhood.

[0037] Step S2234, calculate the difference between the data quantity of the target determined neighborhood and the data quantity of the first monitoring data sequence, and take the difference between the calculation result and 1 as the first data uncertainty coefficient. Specifically, the original data quantity of the first monitoring data sequence is 7, the data quantity of the target determined neighborhood is 6, and the difference between the two is 6-7=-1. Then, the difference between the calculation result (-1) and 1 is 1-(-1)=2, which is the first data uncertainty coefficient. If the data quantity of the target determined neighborhood and the original data quantity are the same, for example, both are 7, the difference is 7-7=0, and the difference between 1 and 0 is 1-0=1, at this time, the coefficient is 1, which means that all original data are included in the target determined neighborhood, and the data uncertainty is very low.

[0038] Step S300, call the pre-constructed digital twin model of chemical production waste gas treatment, combine the plurality of centralized monitoring data arrays and the plurality of data uncertainty coefficient arrays to perform double uncertainty integration from two dimensions of semantic uncertainty and working condition uncertainty, and obtain integrated chemical production waste gas recovery working condition characteristics.

[0039] Specifically, the digital twin model pre-built and 1:1 mapped with the entity system is called, and the centralized monitoring data and uncertainty coefficients obtained in S200 are input into the model. Semantic uncertainty refers to the ambiguity or interpretation difference of the data itself, such as the measurement deviation of different sensors on the same parameter, the interpretation error caused by inconsistent data units or precision, and the unknown fluctuation of the system running state, such as the deviation of the working condition caused by the change of raw material composition and the aging of equipment. Through model calculation, the uncertainty coefficients of the two dimensions are calculated, and then the two are weighted and integrated, and finally the working condition characteristics that can fully reflect the current system running state are output, for example, the current acid gas washing efficiency is 85%, including semantic uncertainty coefficient 0.03, working condition uncertainty coefficient 0.05, and integrated uncertainty coefficient 0.04; the current halogenated aromatic hydrocarbon recovery rate is 78%, including semantic uncertainty coefficient 0.02, working condition uncertainty coefficient 0.06, and integrated uncertainty coefficient 0.045. These characteristics contain parameter values and corresponding uncertainty information, which can provide accurate basis for subsequent adjustment.

[0040] In one possible implementation, a pre-built chemical production waste gas treatment digital twin model is called, and step S300 further includes step S310 of acquiring the process flow chain and key equipment structure of the acid gas washing unit, multi-stage condensing unit and activated carbon adsorption regeneration unit in the chemical production waste gas treatment system, calling a digital twin simulator for twin simulation, and obtaining a chemical production waste gas treatment digital twin model; wherein the chemical production waste gas treatment digital twin model includes: a two-stage acid gas washing tower mass transfer model for simulating the acid gas removal process; a multi-stage low-temperature condensing model for simulating the halogenated aromatic hydrocarbon condensation and capture process; an activated carbon adsorption-desorption bed axial model, a mass transfer model and a heat release model.

[0041] Specifically, a digital mirror image of the entity system is constructed, and first, process and equipment data of the entity system are collected, wherein the process flow chain includes the complete process of waste gas entering the acid gas washing unit, first-stage washing tower, second-stage washing tower, entering the multi-stage condensing unit, first-stage condensing, gas-liquid separation, second-stage condensing, and finally entering the activated carbon adsorption regeneration unit, adsorption tower adsorption, desorption heating, and regeneration cycle. The key equipment structure includes the diameter, height, filler type and height of the washing tower, the heat exchange area of the condenser, the cooling medium type, the bed height of the adsorption tower, the activated carbon particle size and other parameters. A digital twin simulator such as ANSYS, Unity or Aspen Plus is called, and the above data is input to build three core sub-models.

[0042] The two-stage acid gas washing tower mass transfer model is based on the double membrane theory, simulates the mass transfer process of acid gas and washing liquid in the filler layer, and the input parameters include washing liquid flow, concentration, pH value, waste gas flow, acid gas concentration, and the output acid gas removal efficiency and outlet acid gas concentration. The multi-stage low-temperature condensation model is based on the phase equilibrium theory, simulates the condensation process of halogenated aromatic hydrocarbons at different temperatures and pressures, and the input parameters include condensation temperature, pressure, waste gas flow, halogenated aromatic hydrocarbon concentration, and the output condensation efficiency and liquid phase recovery rate. The bed axial model in the activated carbon adsorption-desorption model simulates the activated carbon adsorption amount distribution at different heights in the adsorption tower, the mass transfer model simulates the diffusion process of halogenated aromatic hydrocarbons in the activated carbon particles, and the heat release model simulates the heat release rate and temperature change in the adsorption process. The input parameters include waste gas flow, concentration, adsorption temperature, and desorption heating temperature, and the output adsorption efficiency and desorption regeneration efficiency. The three sub-models are connected in series according to the process flow chain to form a complete digital twin model.

[0043] In one possible implementation, a pre-constructed chemical production waste gas treatment digital twin model is called, and the multiple centralized monitoring data arrays and multiple data uncertainty coefficient arrays are combined to perform double uncertainty integration from the aspects of semantic uncertainty and working condition uncertainty, to obtain integrated chemical production waste gas recovery working condition characteristics. Step S300 further includes step S320 of synchronously transmitting the multiple centralized monitoring data arrays to the chemical production waste gas treatment digital twin model for twin simulation prediction, to obtain digital twin simulation prediction working condition characteristics. Specifically, the centralized monitoring data arrays obtained in S200, such as the pH value of 7.5 and the HCl concentration of 8 mg / m 3 at the outlet of the washing tower, the condenser outlet temperature of 5°C and the pressure of 0.8 MPa, and the VOCs concentration of 20 mg / m 3 at the outlet of the adsorption tower, are synchronously transmitted to the digital twin model according to the time stamp, to ensure that the model input is consistent with the current state of the entity system. Based on the input data, the model simulates the running process of the entity system through internal mass transfer, heat transfer, phase equilibrium calculation, and the like, and predicts the key performance indicators under the current working condition.

[0044] For example, the model calculates the acid gas removal efficiency to be 92% according to the acid gas concentration at the inlet of the washing tower and the monitored concentration at the outlet; calculates the halogenated aromatic hydrocarbon condensation efficiency to be 75% according to the condenser temperature and pressure; and calculates the adsorption efficiency to be 88% according to the VOCs concentration at the inlet and outlet of the adsorption tower. These efficiency indicators, inlet and outlet parameters of each unit, and equipment running parameters jointly constitute the digital twin simulation prediction working condition characteristics, which reflect the current running state of the system predicted by the model.

[0045] Step S330, combine the plurality of data uncertainty coefficient arrays and the digital twin simulation predicted working condition characteristics to perform semantic uncertainty and working condition uncertainty analysis, and obtain semantic uncertainty coefficients and working condition uncertainty coefficients. Specifically, the semantic uncertainty analysis is based on the data uncertainty coefficient arrays, for example, the monitoring data uncertainty coefficients of different sensors for the same parameter are 0.03, 0.04, and 0.02 respectively, and the semantic uncertainty coefficient of the parameter is calculated by weighted average, for example, the high-precision sensor weight is 0.4, the medium-precision is 0.3, and the low-precision is 0.3, and the semantic uncertainty coefficient of the parameter is 0.03×0.4+0.04×0.3+0.02×0.3=0.03. The working condition uncertainty analysis is based on the comparison between the simulation predicted working condition characteristics and the historical data, for example, the current simulation predicted adsorption efficiency is 88%, the number of abnormal scenes similar to the efficiency value is counted by querying the historical abnormal scene library, if there are 20 abnormal scenes in the range of 100 abnormal scenes in the historical library, then the working condition uncertainty coefficient is 20÷100=0.2. Finally, the semantic uncertainty coefficients and the working condition uncertainty coefficients corresponding to each key parameter are output.

[0046] Step S340, based on the semantic uncertainty coefficients and the working condition uncertainty coefficients, weighting is performed to obtain integrated uncertainty coefficients, and the integrated uncertainty coefficients are used to identify the digital twin simulation predicted working condition characteristics to obtain integrated chemical production waste gas recovery working condition characteristics. Specifically, weights are set for the semantic uncertainty coefficients and the working condition uncertainty coefficients, and the weights are determined according to the influence of the two on system operation, for example, the semantic uncertainty affects data reliability, and the weight is 0.4; the working condition uncertainty affects state judgment, and the weight is 0.6. Taking the HCl concentration parameter as an example, the semantic uncertainty coefficient is 0.046, the working condition uncertainty coefficient is 0.075, and the integrated uncertainty coefficient is calculated by weighting: 0.046×0.4+0.075×0.6=0.0184+0.045=0.0634. The integrated uncertainty coefficient is associated with the corresponding parameter in the digital twin simulation predicted working condition characteristics, for example, the HCl concentration is 8 mg / m 3 (integrated uncertainty coefficient 0.0634), the adsorption efficiency is 88% (integrated uncertainty coefficient 0.08), and the condensation efficiency is 75% (integrated uncertainty coefficient 0.05). All the identified parameters jointly constitute the integrated chemical production waste gas recovery working condition characteristics, and comprehensively reflect the parameter values and uncertainty degrees of the current working condition.

[0047] In one possible implementation, step S330 further includes step S331, combining the plurality of data uncertainty coefficient arrays to obtain semantic uncertainty coefficients. Specifically, weights are set for the uncertainty coefficients of each sensor array, and the weights are determined according to the accuracy level of the sensor, the importance of the installation position, and the reliability of the historical data, for example, the HCl concentration sensor at the outlet of the first-stage washing tower has an accuracy of ±1 mg / m3 The weight is set to 0.5, the HCl concentration sensor accuracy of the outlet of the secondary washing tower is ±2 mg / m 3 The weight is set to 0.3, the HCl concentration sensor accuracy of the outlet of the gas-liquid separator is ±3 mg / m 3 The weight is set to 0.2. The coefficients related to the HCl concentration in the plurality of data uncertainty coefficient arrays are 0.03 (primary outlet), 0.05 (secondary outlet), and 0.08 (separator outlet), respectively; by weighted calculation: 0.03*0.5+0.05*0.3+0.08*0.2=0.015+0.015+0.016=0.046, the value is the semantic uncertainty coefficient of the HCl concentration parameter, reflecting the semantic ambiguity degree after the multi-sensor data fusion.

[0048] In step S332, the digital twin simulation predicted working condition characteristics are matched with a historical abnormal chemical production waste gas recovery scene library, and a plurality of matching historical abnormal chemical production waste gas recovery scenes with a matching similarity greater than or equal to a preset matching similarity threshold value are extracted. Specifically, the historical abnormal scene library contains various abnormal working condition data that have occurred in the past system, such as an acid gas washing efficiency lower than 80%, a condensation efficiency lower than 70%, an adsorption efficiency lower than 85%, etc. Each scene contains range values of key parameters. The preset matching similarity threshold value can be set to 80%, that is, if the parameter coincidence degree of the current working condition and the historical scene is ≥80%, it is determined to be matched. The working condition characteristics predicted by the digital twin simulation are compared with the scenes in the historical library, for example, the matching similarity with the historical scene with an adsorption efficiency of 85%-90%, a condensation efficiency of 70%-80%, and a washing efficiency of 80%-90% is 85%≥80%, and the scene is extracted as a matching historical abnormal scene. If there are a plurality of similar scenes, they are all included in the matching result.

[0049] In step S333, the ratio of the number of the plurality of matching historical abnormal chemical production waste gas recovery scenes to the total number of historical abnormal chemical production waste gas recovery scenes in the library is calculated to obtain a working condition uncertainty coefficient. Specifically, assuming that the total number of historical abnormal chemical production waste gas recovery scenes in the library is 200, and the number of matching historical abnormal scenes extracted through S332 is 15, the ratio of the two is calculated: 15÷200=0.075, which is the working condition uncertainty coefficient. The coefficient of 0.075 indicates that the similarity degree of the current working condition and the historical abnormal scene is low, and the uncertainty of the abnormality is low. If the number of matching scenes is 50 and the ratio is 0.25, it indicates that the current working condition is similar to more historical abnormal scenes, the working condition uncertainty is high, and attention should be paid.

[0050] Step S400, according to the integrated chemical production waste gas recovery working condition characteristics, the real-time waste gas recovery process parameters are adaptively adjusted, the adjusted waste gas recovery process parameters are obtained, and the chemical production waste gas treatment system is adjusted based on the adjusted waste gas recovery process parameters.

[0051] Specifically, according to the integrated working condition characteristics, the optimization direction of the current process parameters is determined, for example, if the condensation efficiency is low, it is determined that the condensation temperature, condensation pressure and other parameters need to be adjusted. Then, combined with the integrated uncertainty coefficient, the adjustment range is determined. The smaller the uncertainty coefficient is, the larger the adjustment range can be appropriately increased. The larger the coefficient is, the adjustment range needs to be reduced to avoid system fluctuations. The specific adjustment parameter value is calculated. Then, the feasibility of the adjustment parameter is verified. If the verification is passed, the adjustment parameters are issued to the control system of the entity system to adjust the equipment operating parameters, realize process optimization, for example, by adjusting the refrigerating capacity of the condenser to reach the target temperature, adjusting the inlet flow of the adsorption tower to match the current adsorption efficiency.

[0052] In one possible implementation, step S400 further includes step S410 of extracting the integrated uncertainty coefficient of the integrated chemical production waste gas recovery working condition characteristics, comparing the integrated uncertainty coefficient with a standard uncertainty coefficient, and determining a single adjustment bandwidth. Specifically, the standard uncertainty coefficient is a threshold value set in advance to determine whether the system needs to be adjusted and the adjustment range, for example, ≤0.05 is set as a low threshold value, >0.05 and ≤0.1 is set as a medium threshold value, and >0.1 is set as a high threshold value. The integrated uncertainty coefficient in the integrated working condition characteristics, such as the integrated uncertainty coefficient 0.08 of the adsorption efficiency, is compared with the standard threshold value, 0.05 < 0.08 < 0.1, and it is determined that the uncertainty is medium. According to the preset adjustment bandwidth rule, such as low uncertainty: adjustment bandwidth ±10%, medium: ±5%, and high: ±2%, the single adjustment bandwidth of the process parameter corresponding to the adsorption efficiency is determined to be ±5%. If the parameter is temperature, such as condensation temperature 10℃, the adjustment bandwidth corresponding to medium uncertainty is ±0.5℃, and the single adjustment range is 9.5℃-10.5℃.

[0053] Step S420, based on the single adjustment bandwidth, the real-time waste gas recovery process parameters are adaptively adjusted, and the stage-adjusted waste gas recovery process parameters are obtained. Specifically, the real-time process parameters that need to be adjusted are determined, for example, according to the integrated working condition characteristics, the adsorption efficiency 88% is lower than the target value 90%, and the integrated uncertainty coefficient 0.08 is medium, the inlet flow of the adsorption tower needs to be adjusted. The current real-time flow is 1000m 3 / h, the single adjustment bandwidth is ±5%, the inlet flow needs to be reduced to increase the adsorption efficiency due to the low adsorption efficiency, so the adjustment direction is to reduce 5%. The stage-adjusted waste gas recovery process parameters are calculated: 1000m 3 / h x (1-5%) = 950m 3 / h. After all the parameters that need to be adjusted are calculated, the phase-adjusted waste gas recovery process parameter set is formed.

[0054] In step S430, the phase-adjusted waste gas recovery process parameters are adaptively simulated and verified by using the chemical production waste gas treatment digital twin model. If the verification is passed, the phase-adjusted waste gas recovery process parameters are used as the adjusted waste gas recovery process parameters. Specifically, the phase-adjusted process parameters are input into the digital twin model, and the running state of the system under the parameters is simulated. The model outputs the simulation results. For example, the adsorption efficiency is improved to 91%, reaching the target value of 90%, the system energy consumption does not increase significantly, and is still within a reasonable range, and there is no risk of equipment over-temperature and over-pressure. Therefore, it is determined that the verification is passed, and the phase-adjusted parameters are the final adjusted waste gas recovery process parameters. If the simulation results show that the adsorption efficiency is only improved to 89% and does not reach the target, the verification is not passed, and the single adjustment bandwidth needs to be adjusted again. The phase-adjusted parameters are calculated again and simulated and verified until the verification is passed.

[0055] The embodiments of the present application adopt a chemical waste gas treatment system including acid gas washing, multi-stage condensation, and activated carbon adsorption regeneration units. Multidimensional sensors are arranged at key equipment to construct a sensor array. Each array traverses and collects multi-source waste gas process data in a preset monitoring window and completes uncertainty preprocessing to obtain a centralized monitoring data array and a data uncertainty coefficient array. A pre-constructed digital twin model of the chemical waste gas treatment is called, and working condition characteristics are obtained from the semantic and working condition dual uncertainty dimensions by integrating the two types of arrays. The real-time process parameters are adaptively adjusted according to the characteristics, and the adjusted parameters are applied to the actual system to complete process parameter adjustment and other technical means. The technical problems of insufficient process adjustment accuracy, resulting in fluctuation of recovery efficiency and instability of emission quality in existing chemical production waste gas recovery are solved. The technical effects of improving process adjustment accuracy, improving recovery efficiency, and ensuring emission quality are achieved.

[0056] In the foregoing, reference is made to Figure 1 The digital-twin-based intelligent chemical production waste gas recovery method according to the embodiments of the present application is described in detail. Next, the digital-twin-based intelligent chemical production waste gas recovery system according to the embodiments of the present application will be described with reference to Figure 2 The digital-twin-based intelligent chemical production waste gas recovery system according to the embodiments of the present application is described.

[0057] The digital-twin-based intelligent chemical production waste gas recovery system according to the embodiment of the present application is used to solve the technical problem of insufficient process adjustment precision of the existing chemical production waste gas recovery, which leads to fluctuation of recovery efficiency and instability of emission quality, so as to improve the process adjustment precision, improve the recovery efficiency and guarantee the emission quality. The digital-twin-based intelligent chemical production waste gas recovery system comprises a sensor arrangement module 10, a data acquisition module 20, a double uncertainty integration module 30 and a recovery process parameter adjustment module 40.

[0058] The sensor arrangement module 10 is used to obtain a plurality of key devices of a chemical production waste gas treatment system, arrange multi-dimensional sensors at the plurality of key devices, and obtain a plurality of sensor arrays, wherein the chemical production waste gas treatment system comprises an acid gas washing unit, a multi-stage condensing unit and an activated carbon adsorption regeneration unit; the data acquisition module 20 is used to traverse the plurality of sensor arrays to perform multi-source waste gas process data acquisition and uncertainty preprocessing within a preset monitoring window, to obtain a plurality of centralized monitoring data arrays and a plurality of data uncertainty coefficient arrays; the double uncertainty integration module 30 is used to call a pre-constructed digital twin model of the chemical production waste gas treatment, and combine the plurality of centralized monitoring data arrays and the plurality of data uncertainty coefficient arrays to perform double uncertainty integration from two dimensions of semantic uncertainty and working condition uncertainty, to obtain integrated chemical production waste gas recovery working condition characteristics; and the recovery process parameter adjustment module 40 is used to adaptively adjust real-time waste gas recovery process parameters according to the integrated chemical production waste gas recovery working condition characteristics, to obtain adjusted waste gas recovery process parameters, and adjust the recovery process parameters of the chemical production waste gas treatment system based on the adjusted waste gas recovery process parameters.

[0059] The data acquisition module 20 is specifically configured as follows: as described above, the plurality of sensor arrays are traversed to perform multi-source waste gas process data acquisition and uncertainty preprocessing within a preset monitoring window, to obtain a plurality of centralized monitoring data arrays and a plurality of data uncertainty coefficient arrays. The data acquisition module 20 can further comprise: a process data acquisition unit for traversing the plurality of sensor arrays to perform multi-source waste gas process data acquisition within a preset monitoring window, to obtain a plurality of monitoring data sequence arrays; an uncertainty preprocessing unit for extracting a first monitoring data sequence from the plurality of monitoring data sequence arrays, and performing uncertainty preprocessing on the first monitoring data sequence, to obtain first centralized monitoring data and a first data uncertainty coefficient, wherein the uncertainty preprocessing comprises data centralization screening and uncertainty analysis; and a data adding unit for adding the first centralized monitoring data and the first data uncertainty coefficient into the plurality of centralized monitoring data arrays and the plurality of data uncertainty coefficient arrays.

[0060] The first monitoring data sequence in the plurality of monitoring data sequence arrays is extracted, and the first monitoring data sequence is subjected to uncertainty preprocessing to obtain first centralized monitoring data and a first data uncertainty coefficient. The uncertainty preprocessing includes data centralization screening and uncertainty analysis. The uncertainty preprocessing unit can further include: a mean calculation subunit configured to calculate a data mean of the first monitoring data sequence to obtain an initial data centralization screening center; a mean shift screening subunit configured to perform mean shift screening on the first monitoring data sequence with the initial data centralization screening center as a starting point to determine the first centralized monitoring data; and an uncertainty analysis subunit configured to perform uncertainty analysis on the first monitoring data sequence based on the first centralized monitoring data to determine the first data uncertainty coefficient.

[0061] The first monitoring data sequence is subjected to uncertainty analysis based on the first centralized monitoring data to determine the first data uncertainty coefficient. The uncertainty analysis subunit can further include: a data screening component configured to perform data screening on the first monitoring data sequence according to a preset similarity bandwidth to construct an initial certainty neighborhood of the first centralized monitoring data, wherein the initial certainty neighborhood includes monitoring data in the first monitoring data sequence that has a similarity to the first centralized monitoring data within a preset similarity bandwidth range; a diffusion component configured to diffuse edges of the initial certainty neighborhood by 1 / 2 of the preset similarity bandwidth to obtain a diffusion certainty neighborhood; a judgment component configured to judge whether a neighborhood data amount difference between the diffusion certainty neighborhood and the initial certainty neighborhood is greater than or equal to a preset data amount difference threshold value, and if so, the initial certainty neighborhood is taken as a target certainty neighborhood; and a first data uncertainty coefficient calculation component configured to calculate a data amount difference between the target certainty neighborhood and the first monitoring data sequence, and take a difference between the calculation result and 1 as the first data uncertainty coefficient.

[0062] The judgment component can further include: if the neighborhood data amount difference between the diffusion certainty neighborhood and the initial certainty neighborhood is less than the preset data amount difference threshold value, the edges of the diffusion certainty neighborhood are diffused by 1 / 2 of the preset similarity bandwidth to obtain an iterative diffusion certainty neighborhood, and this process is repeated until a neighborhood data amount difference between two adjacent diffusion processes is greater than or equal to the preset data amount difference threshold value to obtain the target certainty neighborhood.

[0063] The detailed description of the specific configuration of the double uncertainty integration module 30 is explained as follows: As described above, the pre-built chemical production waste gas treatment digital twin model is called, and the double uncertainty integration module 30 can further include: a twin simulation unit for obtaining the process flow chain and key equipment structure of the acid gas washing unit, multi-stage condensation unit and activated carbon adsorption regeneration unit in the chemical production waste gas treatment system, calling the digital twin simulator for twin simulation, and obtaining the chemical production waste gas treatment digital twin model; wherein the chemical production waste gas treatment digital twin model includes: a two-stage acid gas washing tower mass transfer model for simulating the acid gas removal process; a multi-stage low-temperature condensation model for simulating the condensation and capture process of halogenated aromatic hydrocarbon substances; an activated carbon adsorption-desorption bed axial model, a mass transfer model and a heat release model.

[0064] Wherein, the pre-built chemical production waste gas treatment digital twin model is called, and the double uncertainty integration is carried out from the two dimensions of semantic uncertainty and working condition uncertainty by combining the multiple centralized monitoring data arrays and multiple data uncertainty coefficient arrays to obtain the integrated chemical production waste gas recovery working condition characteristics, and the double uncertainty integration module 30 can further include: a twin simulation prediction unit for synchronously transmitting the multiple centralized monitoring data arrays to the chemical production waste gas treatment digital twin model for twin simulation prediction to obtain the digital twin simulation prediction working condition characteristics; an uncertainty analysis unit for combining the multiple data uncertainty coefficient arrays and the digital twin simulation prediction working condition characteristics to perform semantic uncertainty and working condition uncertainty analysis, obtaining the semantic uncertainty coefficient and the working condition uncertainty coefficient; an uncertainty coefficient weighting unit for weighting based on the semantic uncertainty coefficient and the working condition uncertainty coefficient to obtain the integrated uncertainty coefficient, and using the integrated uncertainty coefficient to identify the digital twin simulation prediction working condition characteristics to obtain the integrated chemical production waste gas recovery working condition characteristics.

[0065] Wherein, the uncertainty analysis unit can further include: a semantic uncertainty coefficient acquisition sub-unit for combining the multiple data uncertainty coefficient arrays to obtain the semantic uncertainty coefficient; a matching sub-unit for matching the digital twin simulation prediction working condition characteristics with the historical abnormal chemical production waste gas recovery scene library to extract multiple matching historical abnormal chemical production waste gas recovery scenes with a matching similarity greater than or equal to a preset matching similarity threshold; a working condition uncertainty coefficient acquisition sub-unit for calculating the ratio of the number of the multiple matching historical abnormal chemical production waste gas recovery scenes to the total number of scenes in the historical abnormal chemical production waste gas recovery scene library to obtain the working condition uncertainty coefficient.

[0066] The specific configuration of the recovery process parameter adjustment module 40 is described in detail as follows: as described above, the recovery process parameter adjustment module 40 can further include: a single adjustment bandwidth determination unit for extracting the integrated uncertainty coefficient of the integrated chemical production waste gas recovery working condition characteristics, comparing the integrated uncertainty coefficient with the standard uncertainty coefficient, and determining the single adjustment bandwidth; an adaptive adjustment unit for adaptively adjusting the real-time waste gas recovery process parameter based on the single adjustment bandwidth to obtain a stage adjustment waste gas recovery process parameter; and an adjusted waste gas recovery process parameter generation unit for adaptively simulating and verifying the stage adjustment waste gas recovery process parameter by using the chemical production waste gas treatment digital twin model, and if the verification is passed, the stage adjustment waste gas recovery process parameter is used as the adjusted waste gas recovery process parameter.

[0067] The digital twin-based intelligent chemical production waste gas recovery system provided in the embodiments of the present application can execute the digital twin-based intelligent chemical production waste gas recovery method provided in any of the embodiments of the present application, and has the corresponding function modules and beneficial effects of the execution method.

[0068] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, however, any number of different modules can be used and run on the user terminal and / or server, and each unit and module included is only divided according to the functional logic, but is not limited to the above division, as long as the corresponding functions can be realized; in addition, the specific names of each functional unit are only for easy mutual differentiation, and do not limit the protection scope of the present application.

[0069] The above is only a preferred embodiment of the present application, and does not limit the present application in any form. Although the present application has been disclosed as above with the preferred embodiment, however, it is not intended to limit the present application, and any person skilled in the art can make some changes or modifications to the above disclosed technical content to obtain equivalent embodiments with equivalent changes, as long as they do not deviate from the technical solution of the present application. Any modification, equivalent change and modification of the above embodiments according to the technical essence of the present application are still within the scope of the technical solution of the present application.

Claims

1. A digital-twin-based intelligent recovery method for chemical production waste gas, characterized in that, The method comprises: obtaining a plurality of key devices of a chemical production waste gas treatment system, arranging a multi-dimensional sensor at the plurality of key devices, and obtaining a plurality of sensor arrays, wherein the chemical production waste gas treatment system comprises an acid gas washing unit, a multi-stage condensing unit, and an activated carbon adsorption regeneration unit; traversing the plurality of sensor arrays to perform multi-source waste gas process data acquisition and uncertainty preprocessing within a preset monitoring window, to obtain a plurality of centralized monitoring data arrays and a plurality of data uncertainty coefficient arrays; calling a pre-constructed digital twin model of the chemical production waste gas treatment, combining the plurality of centralized monitoring data arrays and the plurality of data uncertainty coefficient arrays to perform double uncertainty integration from two dimensions of semantic uncertainty and working condition uncertainty, and obtaining integrated chemical production waste gas recovery working condition characteristics; according to the integrated chemical production waste gas recovery working condition characteristics, adaptively adjusting real-time waste gas recovery process parameters to obtain adjusted waste gas recovery process parameters, and adjusting the recovery process parameters of the chemical production waste gas treatment system based on the adjusted waste gas recovery process parameters; traversing the plurality of sensor arrays to perform multi-source waste gas process data acquisition and uncertainty preprocessing within a preset monitoring window, to obtain a plurality of centralized monitoring data arrays and a plurality of data uncertainty coefficient arrays, comprising: traversing the plurality of sensor arrays to perform multi-source waste gas process data acquisition within a preset monitoring window, to obtain a plurality of monitoring data sequence arrays; extracting a first monitoring data sequence from the plurality of monitoring data sequence arrays, performing uncertainty preprocessing on the first monitoring data sequence, and obtaining first centralized monitoring data and a first data uncertainty coefficient, wherein the uncertainty preprocessing comprises data centralization screening and uncertainty analysis; adding the first centralized monitoring data and the first data uncertainty coefficient to the plurality of centralized monitoring data arrays and the plurality of data uncertainty coefficient arrays; extracting a first monitoring data sequence from the plurality of monitoring data sequence arrays, performing uncertainty preprocessing on the first monitoring data sequence, and obtaining first centralized monitoring data and a first data uncertainty coefficient, wherein the uncertainty preprocessing comprises data centralization screening and uncertainty analysis, comprising: calculating the data mean value of the first monitoring data sequence to obtain an initial data centralization center; taking the initial data centralization center as a starting point, performing mean shift screening on the first monitoring data sequence to determine the first centralized monitoring data; based on the first centralized monitoring data, performing uncertainty analysis on the first monitoring data sequence to determine the first data uncertainty coefficient; based on the first centralized monitoring data, performing uncertainty analysis on the first monitoring data sequence to determine the first data uncertainty coefficient, comprising: performing data screening on the first monitoring data sequence according to a preset similarity bandwidth to construct an initial certainty neighborhood of the first centralized monitoring data, wherein the initial certainty neighborhood comprises monitoring data in the first monitoring data sequence that has a similarity to the first centralized monitoring data within a preset similarity bandwidth range; Diffuse edges of the initial deterministic neighborhood by 1 / 2 of preset similarity bandwidth to obtain a diffusion deterministic neighborhood; Determine whether a difference between a data amount of the diffusion deterministic neighborhood and a data amount of the initial deterministic neighborhood is greater than or equal to a preset data amount difference threshold value, and if yes, take the initial deterministic neighborhood as a target deterministic neighborhood; Calculate a difference between the data amount of the target deterministic neighborhood and a data amount of the first monitoring data sequence, and take a difference between a calculation result and 1 as a first data uncertainty coefficient; Call a pre-constructed digital twin model of chemical production waste gas treatment, combine the plurality of centralized monitoring data arrays and the plurality of data uncertainty coefficient arrays to perform double uncertainty integration from two dimensions of semantic uncertainty and working condition uncertainty, and obtain integrated chemical production waste gas recovery working condition characteristics, including: Synchronously transmit the plurality of centralized monitoring data arrays to the digital twin model of chemical production waste gas treatment to perform twin simulation prediction, and obtain digital twin simulation predicted working condition characteristics; Combine the plurality of data uncertainty coefficient arrays and the digital twin simulation predicted working condition characteristics to perform semantic uncertainty and working condition uncertainty analysis, and obtain a semantic uncertainty coefficient and a working condition uncertainty coefficient; Based on the semantic uncertainty coefficient and the working condition uncertainty coefficient, weight is obtained, and the integrated uncertainty coefficient is used to identify the digital twin simulation predicted working condition characteristics to obtain the integrated chemical production waste gas recovery working condition characteristics; Combine the plurality of data uncertainty coefficient arrays to obtain a semantic uncertainty coefficient; Match the digital twin simulation predicted working condition characteristics with a historical abnormal chemical production waste gas recovery scene library, and extract a plurality of matching historical abnormal chemical production waste gas recovery scenes with a matching similarity greater than or equal to a preset matching similarity threshold value; Calculate a ratio between a quantity of the plurality of matching historical abnormal chemical production waste gas recovery scenes and a total scene number in the historical abnormal chemical production waste gas recovery scene library to obtain a working condition uncertainty coefficient.

2. The digital-twin-based intelligent recovery method for chemical production waste gas according to claim 1, characterized in that, If the difference between the data amount of the diffusion deterministic neighborhood and the data amount of the initial deterministic neighborhood is less than the preset data amount difference threshold value, diffuse edges of the diffusion deterministic neighborhood by 1 / 2 of the preset similarity bandwidth to obtain an iterative diffusion deterministic neighborhood, and the process is repeated until a difference between data amounts of two adjacent diffusion neighborhoods is greater than or equal to the preset data amount difference threshold value, and a target deterministic neighborhood is obtained.

3. The digital-twin-based intelligent recovery method of chemical production waste gas according to claim 1, characterized in that, Call a pre-constructed digital twin model of chemical production waste gas treatment, including: Obtain a process flow chain and a key equipment structure of an acid gas washing unit, a multi-stage condensing unit and an activated carbon adsorption regeneration unit in a chemical production waste gas treatment system, call a digital twin simulator to perform twin simulation, and obtain a digital twin model of chemical production waste gas treatment; The digital twin model of chemical production waste gas treatment includes: A two-stage acid gas washing tower mass transfer model for simulating an acid gas removal process; A multi-stage low-temperature condensing model for simulating a halogenated aromatic compound condensation and capture process; An activated carbon adsorption-desorption bed axial model, a mass transfer model and a heat release model.

4. The digital-twin-based intelligent recovery method of chemical production waste gas according to claim 1, characterized in that, including: extracting the integrated uncertainty coefficient of the integrated chemical production waste gas recovery working condition characteristics, comparing the integrated uncertainty coefficient with the standard uncertainty coefficient, and determining the single adjustment bandwidth; based on the single adjustment bandwidth, adaptively adjusting the real-time waste gas recovery process parameters to obtain the stage-adjusted waste gas recovery process parameters; using the chemical production waste gas treatment digital twin model to adaptively simulate and verify the stage-adjusted waste gas recovery process parameters, and if the verification is passed, the stage-adjusted waste gas recovery process parameters are used as the adjusted waste gas recovery process parameters.

5. The intelligent waste gas recycling system for chemical production based on digital twinning, characterized in that, The system is used to implement the digital twin-based intelligent chemical production waste gas recovery method according to any one of claims 1-4, and the system comprises: a sensor arrangement module for obtaining a plurality of key devices of a chemical production waste gas treatment system, arranging a multi-dimensional sensor at the plurality of key devices, and obtaining a plurality of sensor arrays, wherein the chemical production waste gas treatment system comprises an acid gas washing unit, a multi-stage condensing unit, and an activated carbon adsorption regeneration unit; a data acquisition module for traversing the plurality of sensor arrays to perform multi-source waste gas process data acquisition and uncertainty preprocessing within a preset monitoring window, and obtaining a plurality of centralized monitoring data arrays and a plurality of data uncertainty coefficient arrays; a double uncertainty integration module for calling a pre-constructed chemical production waste gas treatment digital twin model, combining the plurality of centralized monitoring data arrays and the plurality of data uncertainty coefficient arrays to perform double uncertainty integration from two dimensions of semantic uncertainty and working condition uncertainty, and obtaining integrated chemical production waste gas recovery working condition characteristics; a recovery process parameter adjustment module for adaptively adjusting real-time waste gas recovery process parameters according to the integrated chemical production waste gas recovery working condition characteristics to obtain adjusted waste gas recovery process parameters, and adjusting the recovery process parameters of the chemical production waste gas treatment system based on the adjusted waste gas recovery process parameters.

Images