Mixing utilization method of coal chemical gasification black water high-temperature flash steam

Abstract

The invention discloses a mixed utilization method of coal chemical gasification black water high-temperature flash steam, and relates to the technical field of black water flash evaporation. In the high-pressure section flash steam heat exchange process, the heat transfer temperature difference and the flow speed are dynamically adjusted through Bayesian optimization, the heat utilization rate of the high-temperature flash steam is maximized, the heat exchange efficiency is improved, and meanwhile in the catalytic reaction stage, the heat transfer temperature difference and the flow speed are dynamically adjusted through Bayesian optimization; heat input and reaction temperature of the catalytic device are regulated in real time based on the outlet temperature of the heat exchanger, so that heat energy distribution is more efficient, and heat waste is avoided; meanwhile, graded flash evaporation is carried out, steam is divided into high-pressure section heat exchange, medium-pressure section catalytic treatment and low-pressure section condensation recovery, and multi-level gradient utilization of energy is achieved; meanwhile, real-time prediction is carried out based on gas components and catalytic device loads, the feeding amount of the adsorbent is dynamically adjusted, resource waste is avoided, meanwhile, the conversion rate of acid gas is increased, and after low-pressure-section tail gas is subjected to condensation treatment, the condensate recovery rate is remarkably increased.

Description

Technical Field

[0001] The invention relates to the technical field of black water flash evaporation, in particular to a method for mixing and utilizing high-temperature flash steam of coal chemical gasification black water. Background Art

[0002] In the field of coal chemical industry, the mixed utilization of black water and high-temperature flash steam is one of the core links of resource recovery and efficient energy conversion. However, the current core equipment, such as heat exchangers, flash tanks and pipelines, have been corroded by high-temperature and high-pressure black water and flash steam for a long time, and are prone to problems such as wear, scaling and corrosion. This increases the frequency and cost of equipment maintenance, and significantly increases the risk of failure under high-load conditions.

[0003] However, equipment failures often occur with lag and randomness. Traditional regular maintenance programs are unable to detect potential hidden dangers in a timely manner, which can easily lead to production risks. In addition, in the black water flash evaporation process, there is a nonlinear coupling relationship between multiple variables such as pressure, temperature, and flow rate. When the pressure in the flash tank fluctuates, traditional control methods cannot quickly adjust to the appropriate parameters, resulting in reduced resource utilization efficiency and even energy waste.

[0004] In response to the above problems, two main countermeasures are currently used: one is to use corrosion-resistant materials and increase the frequency of chemical cleaning to deal with equipment wear and scaling; the other is to adjust process parameters through preset thresholds, but shutdown for cleaning and material replacement will significantly affect production efficiency, and the traditional fixed mode PID controller or preset program responds slowly, making it difficult to meet the real-time requirements of mixed utilization of high-temperature flash steam, and unable to fundamentally solve the above problems. Therefore, a method for mixed utilization of high-temperature flash steam from coal chemical gasification black water is urgently needed to solve such problems. Summary of the invention

[0005] In view of the above existing problems, the present invention is proposed.

[0006] The present invention provides a method for mixed utilization of high-temperature flash steam from coal chemical gasification black water to solve the problem that in traditional mixed utilization schemes, equipment maintenance is mainly based on regular inspections, which is prone to untimely inspections and increases safety risks. In addition, the process parameters of the mixing process are not adjusted in time, and targeted adjustments cannot be achieved, which easily leads to waste of resources.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] The present invention provides a method for mixed utilization of high-temperature flash steam from coal chemical gasification black water, which comprises:

[0009] Step S1, flash data collection,

[0010] Arrange sensors to collect operation data in real time, use long short-term memory network (LSTM) to model the operation data, and predict the equipment operation status trend, including the pressure and gas flow change trend of the flash tank, the scaling trend of the heat exchanger, and the load change of the adsorption device;

[0011] Step S2, graded flash evaporation,

[0012] Based on the pressure change trend of the flash tank in step S1, a graded flash process is set;

[0013] Step S3, high pressure section flash steam heat exchange,

[0014] The flash steam from the high-pressure section enters the heat exchanger to preheat the black water or circulating process water. Combined with the flow and temperature parameters of the high-pressure section, the heat transfer temperature difference and flow rate of the heat exchanger are dynamically adjusted.

[0015] Step S4, catalytic adsorption,

[0016] The flash gas in the medium-pressure section enters the catalytic device to treat the acid gas and recover the organic matter. The reaction temperature and residence time of the catalytic device are dynamically adjusted in combination with the outlet temperature of the heat exchanger.

[0017] Step S5, fault prediction and linkage adjustment,

[0018] Based on the operating data and prediction results of steps S1 to S4, the health status of the flash tank, heat exchanger and adsorption device is evaluated; for equipment predicted to be close to the risk threshold, an early warning is triggered and potential failure points are recorded.

[0019] Furthermore, in step S1, sensors are arranged at the flash tank, heat exchanger, adsorption device and pipeline nodes, and the operating data collected in real time include the pressure, temperature and steam flow of the flash tank, the heat exchange rate, flow rate and inlet and outlet temperature difference of the heat exchanger, and the gas composition, flow rate and equipment operating load of the catalytic adsorption device.

[0020] Furthermore, in step S2, the flash steam in the high-pressure section is introduced into a heat exchanger to provide heat for black water or circulating process water;

[0021] The flash gas in the medium-pressure section is separated into gas and liquid and introduced into a catalytic device to treat acidic gas and recover organic matter;

[0022] The low-pressure tail gas is condensed and then the condensate is recovered for recycling;

[0023] In step S2, according to the pressure and gas flow change trends in step S1 and combined with the actual load demand, the Bayesian optimization model is used to dynamically adjust the steam flow and temperature distribution of each pressure section.

[0024] Furthermore, in step S3, a Bayesian optimization model is used to dynamically adjust the operating parameters of the heat exchanger according to the heat exchanger flow, temperature difference and scaling trend of step S1; a predictive cleaning trigger mechanism is introduced to remind the staff to plan the cleaning time in advance;

[0025] In step S4, based on the gas composition and the load of the catalyst device in step S1, the adsorbent dosage and the catalytic conditions are adjusted in real time, and the thermal balance of the catalyst device is adjusted according to the outlet temperature of the heat exchanger in step S3.

[0026] Furthermore, in step S1, the flash tank pressure and flow rate, heat exchanger fouling trend and adsorption device load change are modeled and predicted:

[0027] To predict the flash tank pressure and flow trend, define the data input matrix as: X t =[P t , F t , T t ], where P t is the pressure of the flash tank at time t, F t is the gas flow rate of the flash tank at time t, T t is the temperature of the flash tank at time t,

[0028] The LSTM state update formula is h t =σ(W h X t +U h h t-1 +b h ), where h t is the hidden state vector at time t, representing the short-term and long-term state memory of the device, W h , U h is the weight matrix of input and hidden state, b h is the bias vector, σ uses Sigmoid and ReLU,

[0029] The output prediction is in, To predict the pressure at time t+k, is the gas flow rate predicted at time t+k, f(·) is the linear regression layer;

[0030] To predict the scaling trend of the heat exchanger, define the data input matrix as Y t =[ΔT t , v t ], where ΔT t is the temperature difference between the heat exchanger inlet and outlet, reflecting the effect of scaling, v t is the fluid flow rate in the heat exchanger,

[0031] The scaling trend model is St+k =αΔT t +βv t +∈, where S t+k is to predict the scaling degree at time t+k, α, β are the weight coefficients of scaling sensitivity to temperature difference and flow rate, ∈ is the noise term,

[0032] To update dynamically: S t+k =g(h t , Y t ), where g(·) is the scaling trend prediction function, which integrates the equipment operation history and current status;

[0033] To predict the load change of the adsorption device, define the data input matrix as Z t =[C t , Q t ], where C t is the concentration of acidic components in the gas at time t, Q t is the air flow rate of the adsorption device at time t,

[0034] The load calculation formula is L t =C t Q t , where L t is the instantaneous load of the adsorption device,

[0035] To forecast load trends: in, is the load of the adsorption device predicted at time t+k, f(·) is the prediction function obtained by LSTM training;

[0036] The prediction of each device status is uniformly modeled as a multi-input multi-output problem, and data splicing is performed:

[0037] D t =[X t , Y t , Z t ], where D t Multi-task prediction for an input matrix that combines flash tank, heat exchanger, and adsorption unit data in, is the prediction result including pressure, flow, scaling trend and load change, and M(·) is the multi-task learning function provided by the trained LSTM network.

[0038] Further, in step S2, graded flash evaporation is performed:

[0039] Using the Bayesian optimization model, the pressure P output from step S1 is synthesized t and gas flow F tThe predicted data and real-time load demand are used to adjust the steam flow and temperature distribution of each pressure section. The objective function is defined as: the distribution of steam in each pressure section meets the needs of the downstream heat exchanger, catalytic device and condensation system, while maximizing the overall efficiency of the system. The objective function is in, To optimize the target value, η h is the thermal efficiency of the high-pressure steam heat exchanger, η c is the treatment efficiency of the medium pressure steam to the catalytic device, η r is the condensation recovery rate of tail gas in the low-pressure section, E is the total energy consumption, including the energy consumption of steam transportation and regulation, and λ is the energy consumption weight coefficient;

[0040] Construct a Bayesian optimization probability model and define the input parameter space as X = {Q h , Q c , Q r}, where Q h is the steam flow rate allocated to the high pressure section, Q c is the steam flow rate allocated to the medium pressure section, Q r is the steam flow rate allocated to the low-pressure section, satisfying Q h +Q c +Q r =F t ;

[0041] The relationship between input and target is constructed using the Gaussian process model: in, is the target distribution probability of a given steam allocation scheme, μ(X) is the mean function, representing the expected performance of the current allocation scheme, and k(X, X′) is the covariance function, measuring the correlation between different allocation schemes;

[0042] Introducing constraints, combined with the pressure trend prediction in step S1, dynamically adjust the upper and lower limits of the allocated pressure to P h ≤P max -ΔP,P r ≥P min +ΔP, where P h is the high pressure section pressure, P r is the low pressure section pressure, P max , P min is the maximum and minimum pressure of the flash tank, and ΔP is the safety margin to ensure stable operation;

[0043] Modeling Steam Temperature Distribution: Among them, T i is the outlet steam temperature of pressure section i, T in is the total steam initial temperature at the flash tank outlet, C p is the specific heat capacity of steam, W iIt is the material flow rate from the flash tank outlet to each section of steam;

[0044] Perform iterative updates of Bayesian optimization and determine the expected improvement value EI as Here is the currently known best target value, and EI(X) is the expected improvement value;

[0045] Dynamically adjust input According to the new X t+1 Allocation plan, calculation of actual target And update the probability model, and obtain the optimal allocation solution through Bayesian optimization:

[0046] Furthermore, in step S3, the heat exchanger operating parameters are dynamically adjusted:

[0047] The Bayesian optimization model is used to dynamically adjust the operating parameters according to the heat exchanger flow, temperature difference and scaling trend in step S1 to construct the heat exchanger performance objective function, which is O = η q -λ e E-λ f S, where is the performance target of the heat exchanger, η q is the heat transfer efficiency, which is defined as the ratio of the actual heat transfer to the ideal heat transfer per unit time, E is the operating energy consumption of the heat exchanger, S is the degree of fouling, and λ e ,λ f are weight coefficients, which respectively represent the influence of energy consumption and scaling on the comprehensive performance;

[0048] Calculate heat transfer efficiency based on the heat transfer equation and actual operating data: in, is the actual heat transfer, equal to UAΔT m , U is the heat transfer coefficient, A is the heat transfer area, ΔT m is the average temperature difference of the heat exchanger, represents the ideal heat transfer, U ideal represents the ideal heat transfer coefficient;

[0049] Here T hot，in , T hot，out are the inlet and outlet temperatures of the hot fluid, T cold，in , T cold，out are the inlet and outlet temperatures of the cold fluid;

[0050] According to the scaling prediction data in step S1, the heat transfer coefficient is dynamically updated to U = U clean ·(1-k s ·S), where U cleanis the initial heat transfer coefficient under non-fouling conditions, k s is the sensitivity coefficient of scaling on the heat transfer coefficient, S is the scaling degree at the current moment;

[0051] Construct a dynamic control model and define the input parameter space as X = {v, ΔT m}, where v is the fluid velocity, ΔT m is the average temperature difference, and the target distribution probability is Here μ(X) is the target mean of the current operating parameters, k(X, X′) is the covariance function of different parameter combinations, and the updated expected improvement value EI is Here Dynamically select the optimal parameters for the currently known best target value

[0052] Furthermore, in step S3, the predictive cleaning trigger mechanism is:

[0053] Define the cleaning trigger threshold as Among them, S threshold is the scaling threshold that triggers cleaning, Δη q / Δt is the rate of decrease of heat transfer efficiency per unit time, τ is the advance time of cleaning triggering;

[0054] Combined with the real-time scaling degree S(t), when S(t)≥S threshold , that is, when the scaling degree S(t) reaches the critical value, the cleaning reminder is triggered.

[0055] Furthermore, in step S4, catalytic regulation is performed based on gas composition and load:

[0056] Adjust the amount of adsorbent, the adsorbent amount calculation formula is: M a =C a Q g ·τ, where M a is the amount of adsorbent required per unit time, C a is the real-time concentration of the target component in the gas, Q g is the gas flow rate of the catalytic device, τ is the residence time of the adsorption reaction;

[0057] The adsorbent consumption rate is dynamically updated to R a (t) = R a (t-1)-M a Δt, where R a (t) is the remaining adsorbent dose at time t, Δt is the time step, when R a (t)≤R min When the supplementary alarm is triggered, R min It is the minimum safety stock;

[0058] The goal of the catalytic reaction is to increase the conversion rate of the acidic components while avoiding the deactivation of the adsorbent due to excessively high temperatures. The conversion rate model is: Among them, η c is the conversion rate of acidic gas, K is the rate constant of catalytic reaction, and the reaction rate constant is expressed by Arrhenius formula Among them, A is the pre-exponential factor, E a is the reaction activation energy, R is the gas constant, T r The reaction temperature of the catalytic device is adjusted in real time according to the actual device requirements. r and τ, so that η c Meet operational objectives.

[0059] Furthermore, in step S4, the thermal balance is adjusted:

[0060] Based on the heat exchanger outlet temperature T provided in step S3 out , dynamically adjust the heat input Q of the catalytic device r Make the system in thermal equilibrium: Q r =Q in -Q loss +Q rxn , where Q r The heat required to maintain thermal balance in the catalytic device, Q in =m g ·C p ·(T out -T in ) is the heat brought into the gas, m g is the gas mass flow rate, C p is the specific heat capacity of the gas, Q loss =hA(T r -T env ) is heat loss, h is heat transfer coefficient, A is equipment surface area, T env is the ambient temperature, Q rxn To release heat in the reaction, adjust Q r and reaction temperature T r , so that the thermal balance is stable and the reaction efficiency is maximized.

[0061] The beneficial effects of the present invention are:

[0062] The present invention dynamically adjusts the heat transfer temperature difference and flow rate through Bayesian optimization during the heat exchange process of the flash steam in the high-pressure section, maximizes the heat utilization rate of the high-temperature flash steam, and improves the heat exchange efficiency. At the same time, in the catalytic reaction stage, the heat input and reaction temperature of the catalytic device are regulated in real time based on the outlet temperature of the heat exchanger, so that the heat energy distribution is more efficient and the heat waste is avoided. At the same time, graded flash evaporation is carried out to divide the steam into high-pressure section heat exchange, medium-pressure section catalytic treatment and low-pressure section condensation recovery, so as to realize multi-level cascade utilization of energy.

[0063] The present invention makes real-time predictions based on gas composition and catalytic device load, dynamically adjusts the amount of adsorbent added, avoids waste of resources, and improves the conversion rate of acidic gas. After the tail gas in the low-pressure section is condensed, the recovery rate of condensate is significantly improved. During the operation of the flash tank, the heat exchanger and the catalytic device, the operating parameters of the equipment are monitored and predicted in real time, and the operating conditions are adjusted dynamically to keep the equipment in an efficient and safe operating state. The graded flash evaporation and heat balance control are coordinated, and the various devices are linked and coordinated to avoid local high efficiency and local low efficiency.

[0064] The present invention, based on the long short-term memory network LSTM prediction model and multi-task learning, conducts real-time evaluation of the health status of the flash tank, heat exchanger and adsorption device, identifies potential operating risks of the equipment in advance, and introduces a predictive cleaning trigger mechanism in the heat exchanger. Based on the dynamic change trend of scaling degree and heat transfer efficiency, the cleaning time is planned to reduce the number of shutdowns and unplanned maintenance. At the same time, during the operation of the heat exchanger, the flow rate and temperature difference parameters are optimized to reduce the energy consumption of steam transportation and circulation. The fault warning and dynamic adjustment mechanism effectively reduce the risk of sudden equipment failure. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0066] Figure 1 The present invention is a schematic flow chart of the method for mixed utilization of high-temperature flash steam from coal chemical gasification black water. DETAILED DESCRIPTION

[0067] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0068] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0069] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0070] Example 1, reference Figure 1 This embodiment provides a method for mixing and utilizing high-temperature flash steam from coal chemical gasification black water, comprising the following steps:

[0071] Step S1, flash data collection,

[0072] Arrange sensors to collect operation data in real time, use long short-term memory network (LSTM) to model the operation data, and predict the equipment operation status trend, including the pressure and gas flow change trend of the flash tank, the scaling trend of the heat exchanger, and the load change of the adsorption device;

[0073] In step S1, sensors are arranged at the flash tank, heat exchanger, adsorption device and pipeline nodes, and the real-time collected operation data include the pressure, temperature and steam flow of the flash tank, the heat exchange rate, flow rate and inlet and outlet temperature difference of the heat exchanger, and the gas composition, flow rate and equipment operation load of the catalytic adsorption device;

[0074] In step S1, the flash tank pressure and flow, heat exchanger fouling trend and adsorption device load change are modeled and predicted:

[0075] To predict the flash tank pressure and flow trend, define the data input matrix as: X t =[P t , F t , T t ], where P t is the pressure of the flash tank at time t, F t is the gas flow rate of the flash tank at time t, T t is the temperature of the flash tank at time t,

[0076] The LSTM state update formula is h t =σ(W h X t +U h h t-1 +b h ), where h t is the hidden state vector at time t, representing the short-term and long-term state memory of the device, W h , U h is the weight matrix of input and hidden state, b h is the bias vector, σ uses Sigmoid and ReLU,

[0077] The output prediction is in, To predict the pressure at time t+k, is the gas flow rate predicted at time t+k, f(·) is the linear regression layer;

[0078] To predict the scaling trend of the heat exchanger, define the data input matrix as Y t =[ΔT t , v t ], where ΔT t is the temperature difference between the heat exchanger inlet and outlet, reflecting the effect of scaling, v t is the fluid flow rate in the heat exchanger,

[0079] The scaling trend model is S t+k =αΔT t +βv t +∈, where S t+k is to predict the scaling degree at time t+k, α, β are the weight coefficients of scaling sensitivity to temperature difference and flow rate, ∈ is the noise term,

[0080] To update dynamically: S t+k =g(h t , Y t ), where g(·) is the scaling trend prediction function, which integrates the equipment operation history and current status;

[0081] To predict the load change of the adsorption device, define the data input matrix as Z t =[C t , Q t ], where C t is the concentration of acidic components in the gas at time t, Q t is the air flow rate of the adsorption device at time t,

[0082] The load calculation formula is L t =C t Q t , where L t is the instantaneous load of the adsorption device,

[0083] To forecast load trends: in, is the load of the adsorption device predicted at time t+k, f(·) is the prediction function obtained by LSTM training;

[0084] The prediction of each device status is uniformly modeled as a multi-input multi-output problem, and data splicing is performed:

[0085] D t =[X t , Y t , Z t ], where D t Multi-task prediction for an input matrix that combines flash tank, heat exchanger, and adsorption unit data in, is the prediction result including pressure, flow, scaling trend and load change, M(·) is the multi-task learning function provided by the trained LSTM network;

[0086] Specifically, based on the characteristics of the specific equipment operating status, the long short-term memory network (LSTM) modeling is used to model and predict the flash tank pressure and flow, heat exchanger fouling trend and adsorption device load changes, providing a basis for subsequent graded flash evaporation, heat exchange regulation and catalytic adsorption.

[0087] Step S2, graded flash evaporation,

[0088] Based on the pressure change trend of the flash tank in step S1, a graded flash process is set;

[0089] In step S2, the flash steam in the high-pressure section is introduced into a heat exchanger to provide heat for black water or circulating process water;

[0090] The flash gas in the medium-pressure section is separated into gas and liquid and introduced into a catalytic device to treat acidic gas and recover organic matter;

[0091] The low-pressure tail gas is condensed and then the condensate is recovered for recycling;

[0092] In step S2, according to the pressure and gas flow rate change trends in step S1 and in combination with the actual load demand, the Bayesian optimization model is used to dynamically adjust the steam flow rate and temperature distribution of each pressure section;

[0093] In step S2, graded flash evaporation is performed:

[0094] Using the Bayesian optimization model, the pressure P output from step S1 is synthesized t and gas flow F t The predicted data and real-time load demand are used to adjust the steam flow and temperature distribution of each pressure section. The objective function is defined as: the distribution of steam in each pressure section meets the needs of the downstream heat exchanger, catalytic device and condensation system, while maximizing the overall efficiency of the system. The objective function is in, To optimize the target value, η h is the thermal efficiency of the high-pressure steam heat exchanger, η c is the treatment efficiency of the medium pressure steam to the catalytic device, η r is the condensation recovery rate of tail gas in the low-pressure section, E is the total energy consumption, including the energy consumption of steam transportation and regulation, and λ is the energy consumption weight coefficient;

[0095] Construct a Bayesian optimization probability model and define the input parameter space as X = {Q h , Q c , Q r}, where Q h is the steam flow rate allocated to the high pressure section, Qc is the steam flow rate allocated to the medium pressure section, Q r is the steam flow rate allocated to the low-pressure section, satisfying Q h +Q c +Q r =F t ;

[0096] The relationship between input and target is constructed using the Gaussian process model: in, is the target distribution probability of a given steam allocation scheme, μ(X) is the mean function, representing the expected performance of the current allocation scheme, and k(X, X′) is the covariance function, measuring the correlation between different allocation schemes;

[0097] Introducing constraints, combined with the pressure trend prediction in step S1, dynamically adjust the upper and lower limits of the allocated pressure to P h ≤P max -ΔP,P r ≥P min +ΔP, where P h is the high pressure section pressure, P r is the low pressure section pressure, P max , P min is the maximum and minimum pressure of the flash tank, and ΔP is the safety margin to ensure stable operation;

[0098] Modeling Steam Temperature Distribution: Among them, T i is the outlet steam temperature of pressure section i, T in is the total steam initial temperature at the flash tank outlet, C p is the specific heat capacity of steam, W i It is the material flow rate from the flash tank outlet to each section of steam;

[0099] Perform iterative updates of Bayesian optimization and determine the expected improvement value EI as Here is the currently known best target value, and EI(X) is the expected improvement value;

[0100] Dynamically adjust input According to the new X t+1 Allocation plan, calculation of actual target And update the probability model, and obtain the optimal allocation solution through Bayesian optimization:

[0101] Specifically, the real-time pressure and flow prediction, energy consumption constraints and equipment load requirements are dynamically combined, and Bayesian optimization is used to gradually approach the optimal allocation plan to ensure that the allocation of steam resources in each work section reaches the global optimal level, and the optimization efficiency is improved through constraint and probability modeling.

[0102] Step S3, high pressure section flash steam heat exchange,

[0103] The flash steam from the high-pressure section enters the heat exchanger to preheat the black water or circulating process water. Combined with the flow and temperature parameters of the high-pressure section, the heat transfer temperature difference and flow rate of the heat exchanger are dynamically adjusted.

[0104] In step S3, the Bayesian optimization model is used to dynamically adjust the operating parameters of the heat exchanger according to the heat exchanger flow, temperature difference and scaling trend of step S1; a predictive cleaning trigger mechanism is introduced to remind the staff to plan the cleaning time in advance;

[0105] In step S3, the heat exchanger operating parameters are dynamically adjusted:

[0106] The Bayesian optimization model is used to dynamically adjust the operating parameters according to the heat exchanger flow, temperature difference and scaling trend in step S1 to construct the heat exchanger performance objective function. The objective function is: in, is the performance target of the heat exchanger, η q is the heat transfer efficiency, which is defined as the ratio of the actual heat transfer to the ideal heat transfer per unit time, E is the operating energy consumption of the heat exchanger, S is the degree of fouling, and λ e ,λ f are weight coefficients, which respectively represent the influence of energy consumption and scaling on the comprehensive performance;

[0107] Calculate heat transfer efficiency based on the heat transfer equation and actual operating data: in, is the actual heat transfer, equal to UAΔT m , U is the heat transfer coefficient, A is the heat transfer area, ΔT m is the average temperature difference of the heat exchanger, represents the ideal heat transfer, U ideal represents the ideal heat transfer coefficient;

[0108] Here T hot，in , T hot，out are the inlet and outlet temperatures of the hot fluid, T cold，in , T cold，out are the inlet and outlet temperatures of the cold fluid;

[0109] According to the scaling prediction data in step S1, the heat transfer coefficient is dynamically updated to U = U clean ·(1-k s ·S), where U clean is the initial heat transfer coefficient under non-fouling conditions, k s is the sensitivity coefficient of scaling on the heat transfer coefficient, S is the scaling degree at the current moment;

[0110] Construct a dynamic control model and define the input parameter space as X = {v, ΔT m}, where v is the fluid velocity, ΔT m is the average temperature difference, and the target distribution probability is Here μ(X) is the target mean of the current operating parameters, k(X, X′) is the covariance function of different parameter combinations, and the updated expected improvement value EI is Here Dynamically select the optimal parameters for the currently known best target value

[0111] In step S3, the predictive cleaning trigger mechanism is:

[0112] Define the cleaning trigger threshold as Among them, S threshold is the scaling threshold that triggers cleaning, Δη q / Δt is the rate of decrease of heat transfer efficiency per unit time, τ is the advance time of cleaning triggering;

[0113] Combined with the real-time scaling degree S(t), when S(t)≥S threshold , that is, when the scaling degree S(t) reaches the critical value, the cleaning reminder is triggered;

[0114] Specifically, the objective function in step S3 comprehensively considers the effects of heat transfer efficiency, energy consumption and scaling, and adjusts the operating parameters in real time through Bayesian optimization. At the same time, it triggers a cleaning reminder in advance according to the rate of decrease in heat transfer efficiency and the degree of scaling.

[0115] Step S4, catalytic adsorption,

[0116] The flash gas in the medium-pressure section enters the catalytic device to treat the acid gas and recover the organic matter. The reaction temperature and residence time of the catalytic device are dynamically adjusted in combination with the outlet temperature of the heat exchanger.

[0117] In step S4, based on the gas composition and the load of the catalyst device in step S1, the amount of adsorbent added and the catalytic conditions are adjusted in real time, and the heat balance of the catalyst device is adjusted according to the outlet temperature of the heat exchanger in step S3;

[0118] In step S4, catalytic regulation is performed based on gas composition and load:

[0119] Adjust the amount of adsorbent, the adsorbent amount calculation formula is: M a =C a Q g ·τ, where M a is the amount of adsorbent required per unit time, C a is the real-time concentration of the target component in the gas, Q g is the gas flow rate of the catalytic device, τ is the residence time of the adsorption reaction;

[0120] The adsorbent consumption rate is dynamically updated to R a (t) = R a (t-1)-M a Δt, where R a (t) is the remaining adsorbent dose at time t, Δt is the time step, when R a (t)≤R min When the supplementary alarm is triggered, R min It is the minimum safety stock;

[0121] The goal of the catalytic reaction is to increase the conversion rate of the acidic components while avoiding the deactivation of the adsorbent due to excessively high temperatures. The conversion rate model is: Among them, η c is the conversion rate of acidic gas, K is the rate constant of catalytic reaction, and the reaction rate constant is expressed by Arrhenius formula Among them, A is the pre-exponential factor, E a is the reaction activation energy, R is the gas constant, T r The reaction temperature of the catalytic device is adjusted in real time according to the actual device requirements. r and τ, so that η c Meeting operational objectives;

[0122] In step S4, the thermal balance is adjusted:

[0123] Based on the heat exchanger outlet temperature T provided in step S3 out , dynamically adjust the heat input Q of the catalytic device r Make the system in thermal equilibrium: Q r =Q in -Q loss +Q rxn , where Q r The heat required to maintain thermal balance in the catalytic device, Q in =m g ·C p ·(T out -T in ) is the heat brought into the gas, m g is the gas mass flow rate, C p is the specific heat capacity of the gas, Q loss =hA(T r -T env ) is heat loss, h is heat transfer coefficient, A is equipment surface area, T env is the ambient temperature, Q rxn To release heat in the reaction, adjust Q r and reaction temperature T r , so that the thermal balance is stable and the reaction efficiency is maximized;

[0124] Specifically, step S4 dynamically optimizes the amount of adsorbent added, catalytic conditions and thermal balance to achieve efficient reaction.

[0125] Step S5, fault prediction and linkage adjustment,

[0126] Based on the operating data and prediction results of steps S1 to S4, the health status of the flash tank, heat exchanger and adsorption device is evaluated; for equipment predicted to be close to the risk threshold, an early warning is triggered and potential failure points are recorded.

[0127] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for mixing and utilizing high-temperature flash steam from coal chemical gasification black water, characterized in that: include, Step S1, flash data collection, Arrange sensors to collect operation data in real time, use long short-term memory network (LSTM) to model the operation data, and predict the equipment operation status trend, including the pressure and gas flow change trend of the flash tank, the scaling trend of the heat exchanger, and the load change of the adsorption device; Step S2, graded flash evaporation, Based on the pressure change trend of the flash tank in step S1, a graded flash distillation process is set; Step S3, high pressure section flash steam heat exchange, The flash steam from the high-pressure section enters the heat exchanger to preheat black water or circulating process water. Combined with the flow and temperature parameters of the high-pressure section, the heat transfer temperature difference and flow rate of the heat exchanger are dynamically adjusted. Step S4, catalytic adsorption, The flash gas from the medium-pressure section enters the catalytic device to treat the acidic gas and recover the organic matter. The reaction temperature and residence time of the catalytic device are dynamically adjusted in combination with the outlet temperature of the heat exchanger. Step S5, fault prediction and linkage adjustment, Based on the operating data and prediction results of steps S1 to S4, the health status of the flash tank, heat exchanger and adsorption device is evaluated; for equipment predicted to be close to the risk threshold, an early warning is triggered and potential failure points are recorded.

2. The method for mixed utilization of high-temperature flash steam from coal chemical gasification black water according to claim 1, characterized in that: In step S1, sensors are arranged at the flash tank, heat exchanger, adsorption device and pipeline nodes, and the operating data collected in real time include the pressure, temperature and steam flow of the flash tank, the heat exchange rate, flow rate and inlet and outlet temperature difference of the heat exchanger, and the gas composition, flow rate and equipment operating load of the catalytic adsorption device.

3. The method for mixed utilization of high-temperature flash steam from coal chemical gasification black water according to claim 2, characterized in that: In step S2, the flash steam in the high-pressure section is introduced into a heat exchanger to provide heat for black water or circulating process water; The flash gas in the medium-pressure section is separated into gas and liquid and introduced into a catalytic device to treat acidic gas and recover organic matter; The low-pressure tail gas is condensed and then the condensate is recovered for recycling; In step S2, according to the pressure and gas flow change trends in step S1 and combined with the actual load demand, the Bayesian optimization model is used to dynamically adjust the steam flow and temperature distribution of each pressure section.

4. The method for mixed utilization of high-temperature flash steam from coal chemical gasification black water according to claim 3, characterized in that: In step S3, the Bayesian optimization model is used to dynamically adjust the operating parameters of the heat exchanger according to the heat exchanger flow, temperature difference and scaling trend of step S1; a predictive cleaning trigger mechanism is introduced to remind the staff to plan the cleaning time in advance; In step S4, based on the gas composition and the load of the catalyst device in step S1, the adsorbent dosage and the catalytic conditions are adjusted in real time, and the thermal balance of the catalyst device is adjusted according to the outlet temperature of the heat exchanger in step S3.

5. The method for mixed utilization of high-temperature flash steam from coal chemical gasification black water according to claim 4, characterized in that: In step S1, the flash tank pressure and flow, heat exchanger fouling trend and adsorption device load change are modeled and predicted: To predict the flash tank pressure and flow trend, define the data input matrix as: X t =[P t , F t , T t ], where P t is the pressure of the flash tank at time t, F t is the gas flow rate of the flash tank at time t, T t is the temperature of the flash tank at time t, The LSTM state update formula is h t =σ(W h X t +U h h t-1 +b h ), where h t is the hidden state vector at time t, representing the short-term and long-term state memory of the device, W h , U h is the weight matrix of input and hidden state, b h is the bias vector, σ uses Sigmoid and ReLU, The output prediction is in, To predict the pressure at time t+k, is the gas flow rate predicted at time t+k, f(·) is the linear regression layer; To predict the scaling trend of the heat exchanger, define the data input matrix as Y t =[ΔT t , v t ], where ΔT t is the temperature difference between the heat exchanger inlet and outlet, reflecting the effect of scaling, v t is the fluid flow rate in the heat exchanger, The scaling trend model is S t+k =αΔT t +βv t +∈, where S t+k To predict the scaling degree at time t+k, α, β are the weight coefficients of scaling sensitivity to temperature difference and flow rate, ∈ is the noise term, To update dynamically: S t+k =g(h t ,Y t ), where g(·) is the scaling trend prediction function, which integrates the equipment operation history and current status; To predict the load change of the adsorption device, define the data input matrix as Z t =[C t ,Q t ], where C t is the concentration of acidic components in the gas at time t, Q t is the air flow rate of the adsorption device at time t, The load calculation formula is L t =C t Q t , where L t is the instantaneous load of the adsorption device, To forecast load trends: in, is the load of the adsorption device predicted at time t+k, f(·) is the prediction function obtained by LSTM training; The prediction of each device status is uniformly modeled as a multi-input multi-output problem, and data splicing is performed: D t =[X t ,Y t ,Z t ], where D t Multi-task prediction for an input matrix that combines flash tank, heat exchanger, and adsorption unit data in, is the prediction result including pressure, flow, scaling trend and load change, and M(·) is the multi-task learning function provided by the trained LSTM network.

6. The method for mixed utilization of high-temperature flash steam from coal chemical gasification black water according to claim 5, characterized in that: In step S2, graded flash evaporation is performed: Using the Bayesian optimization model, the pressure P output from step S1 is synthesized t and gas flow F t The predicted data and real-time load demand are used to adjust the steam flow and temperature distribution of each pressure section. The objective function is defined as: the distribution of steam in each pressure section meets the needs of the downstream heat exchanger, catalytic device and condensation system, while maximizing the overall efficiency of the system. The objective function is in, To optimize the target value, η h is the thermal efficiency of the high-pressure steam heat exchanger, η c is the treatment efficiency of the medium pressure steam to the catalytic device, η r is the condensation recovery rate of tail gas in the low-pressure section, E is the total energy consumption, including the energy consumption of steam transportation and regulation, and λ is the energy consumption weight coefficient; Construct a Bayesian optimization probability model and define the input parameter space as X = {Q h , Q c , Q r }, where Q h is the steam flow rate allocated to the high pressure section, Q c is the steam flow rate allocated to the medium pressure section, Q r is the steam flow rate allocated to the low-pressure section, satisfying Q h +Q c +Q r =F t ; The relationship between input and target is constructed using the Gaussian process model: in, is the target distribution probability of a given steam allocation scheme, μ(X) is the mean function, representing the expected performance of the current allocation scheme, and k(X, X′) is the covariance function, measuring the correlation between different allocation schemes; Introducing constraints, combined with the pressure trend prediction in step S1, dynamically adjust the upper and lower limits of the allocated pressure to P h ≤P max -ΔP,O r ≥P min +ΔP, where P h is the high pressure section pressure, O r is the low pressure section pressure, P max , P min is the maximum and minimum pressure of the flash tank, and ΔP is the safety margin to ensure stable operation; Modeling Steam Temperature Distribution: Among them, T i is the outlet steam temperature of pressure section i, T in is the total steam initial temperature at the flash tank outlet, C p is the specific heat capacity of steam, W i It is the material flow rate from the flash tank outlet to each section of steam; Perform iterative updates of Bayesian optimization and determine the expected improvement value EI as Here is the currently known best target value, and EI(X) is the expected improvement value; Dynamically adjust input According to the new X t+1 Allocation plan, calculation of actual target And update the probability model, and obtain the optimal allocation solution through Bayesian optimization:

7. The method for mixed utilization of high-temperature flash steam from coal chemical gasification black water according to claim 6, characterized in that: In step S3, the heat exchanger operating parameters are dynamically adjusted: The Bayesian optimization model is used to dynamically adjust the operating parameters according to the heat exchanger flow, temperature difference and scaling trend in step S1 to construct the heat exchanger performance objective function. The objective function is: in, is the performance target of the heat exchanger, η q is the heat transfer efficiency, which is defined as the ratio of the actual heat transfer to the ideal heat transfer per unit time, E is the operating energy consumption of the heat exchanger, S is the degree of fouling, and λ e ,λ f are weight coefficients, which respectively represent the influence of energy consumption and scaling on the comprehensive performance; Calculate heat transfer efficiency based on the heat transfer equation and actual operating data: in, is the actual heat transfer, equal to UAΔT m , U is the heat transfer coefficient, A is the heat transfer area, ΔT m is the average temperature difference of the heat exchanger, represents the ideal heat transfer, U ideal represents the ideal heat transfer coefficient; Here T hot，in , T hot，out are the inlet and outlet temperatures of the hot fluid, T cold，in , T cold，out are the inlet and outlet temperatures of the cold fluid; According to the scaling prediction data in step S1, the heat transfer coefficient is dynamically updated to U = U clean ·(1-k s ·S), where U clean is the initial heat transfer coefficient under non-fouling conditions, k s is the sensitivity coefficient of scaling on the heat transfer coefficient, S is the scaling degree at the current moment; Construct a dynamic control model and define the input parameter space as X = {v, ΔT m }, where v is the fluid velocity, ΔT m is the average temperature difference, and the target distribution probability is Here μ(X) is the target mean of the current operating parameters, k(X, X′) is the covariance function of different parameter combinations, and the updated expected improvement value EI is Here Dynamically select the optimal parameters for the currently known best target value 8. The method for mixed utilization of high-temperature flash steam from coal chemical gasification black water according to claim 7, characterized in that: In step S3, the predictive cleaning trigger mechanism is: Define the cleaning trigger threshold as Among them, S threshold is the scaling threshold that triggers cleaning, Δη q / Δt is the rate of decrease of heat transfer efficiency per unit time, τ is the advance time of cleaning triggering; Combined with the real-time scaling degree S(t), when S(t)≥S threshold , that is, when the scaling degree S(t) reaches the critical value, the cleaning reminder is triggered.

9. The method for mixed utilization of high-temperature flash steam from coal chemical gasification black water according to claim 8, characterized in that: In step S4, catalytic regulation is performed based on gas composition and load: Adjust the amount of adsorbent, the adsorbent amount calculation formula is: M a =C a Q g ·τ, where M a is the amount of adsorbent required per unit time, C a is the real-time concentration of the target component in the gas, Q g is the gas flow rate of the catalytic device, τ is the residence time of the adsorption reaction; The adsorbent consumption rate is dynamically updated to R a (t) = R a (t-1)-M a Δt, where R a (t) is the remaining adsorbent dose at time t, Δt is the time step, when R a (t)≤R min When the supplementary alarm is triggered, R min It is the minimum safety stock; The catalytic reaction conversion model is defined as Among them, η c is the conversion rate of acidic gas, K is the rate constant of catalytic reaction, and the reaction rate constant is expressed by Arrhenius formula Among them, A is the pre-exponential factor, E a is the reaction activation energy, R is the gas constant, T r The reaction temperature of the catalytic device is adjusted in real time according to the actual device requirements. r and τ, so that η c Meet operational objectives.

10. The method for mixed utilization of high-temperature flash steam from coal chemical gasification black water according to claim 9, characterized in that: In step S4, the thermal balance is adjusted: Based on the heat exchanger outlet temperature T provided in step S3 out , dynamically adjust the heat input Q of the catalytic device r Make the system in thermal equilibrium: Q r =Q in -Q loss +Q rxn , where Q r The heat required to maintain thermal balance in the catalytic device, Q in =m g ·C p ·(T out -T in ) is the heat brought into the gas, m g is the gas mass flow rate, C p is the specific heat capacity of the gas, Q lo s s =hA(T r -T env ) is heat loss, h is heat transfer coefficient, A is equipment surface area, T env is the ambient temperature, Q rxn To release heat in the reaction, adjust Q r and reaction temperature T r , so that the thermal balance is stable and the reaction efficiency is maximized.

Images