A waste heat water production system and method based on membrane distillation

Abstract

The present application relates to waste heat energy recycling technical field, propose a kind of waste heat water making system and method based on membrane distillation, comprising: in waste heat main pipeline arrangement several nodes and collect continuous time's valve port node temperature, and the inlet temperature of each level process, record ambient temperature and waste heat flow;Obtain the inlet predicted temperature of each level process in future several times;Obtain the single-step fitness of each level process in future each time;Obtain the overall fitness of the inlet predicted temperature of each level process in future several windows;Obtain the comprehensive influence degree of the inlet predicted temperature of each level process in future several windows;Combined with the pipeline change relationship between each window and each level process, obtain the comprehensive score of each level process in future each window;Control pipeline switching and run hierarchical membrane distillation water making process.The present application aims at solving the problem that existing membrane distillation waste heat water making technology is difficult to realize waste heat cascade maximization utilization.

Description

Technical Field

[0001] This invention relates to the field of waste heat energy recovery and utilization technology, specifically to a waste heat water production system and method based on membrane distillation. Background Technology

[0002] Driven by the dual demands of industrial energy conservation and carbon reduction and water resource recycling, waste heat water treatment solutions based on membrane distillation technology have become an important technological direction for the utilization of industrial waste heat resources. Among them, the three-stage membrane distillation process of DCMD, VMD, and AGMD can cover the tiered utilization of waste heat in a wide range of 30-100℃, with significant environmental protection and energy-saving benefits. However, common problems in industrial scenarios include large fluctuations in waste heat temperature, natural temperature drop during pipeline transmission, and overlapping temperature adaptation ranges of different membrane processes, which bring core industry challenges to the accurate allocation and full utilization of waste heat.

[0003] Existing membrane distillation waste heat water treatment technologies mostly rely on simple threshold temperature control modes to regulate waste heat temperature and flow direction. The heating stage requires additional electrical and thermal energy consumption, while the cooling stage directly wastes high-quality waste heat. At the same time, they do not make fine-grained flow direction allocation for the uncertainty of waste heat temperature fluctuations, but only use a binary judgment logic of "flowing in when the temperature reaches the standard", resulting in low waste heat utilization efficiency in temperature overlap ranges. Moreover, the lack of quantitative decision-making basis for pipeline switching makes it easy for frequent switching to aggravate valve and membrane module wear, or for switching lag to cause shortened membrane life and large fluctuations in water production efficiency. It is impossible to balance membrane module life, water production efficiency and pipeline switching frequency in scenarios with large waste heat fluctuations and overlapping process temperature ranges, making it difficult to achieve maximum utilization of waste heat and long-term stable operation of the system. Summary of the Invention

[0004] This invention provides a waste heat water production system and method based on membrane distillation, to solve the problem that existing membrane distillation waste heat water production technologies cannot achieve maximum utilization of waste heat in a cascade manner. The specific technical solution adopted is as follows: This invention proposes a waste heat water production method based on membrane distillation, which includes the following steps: Several nodes are arranged in the main waste heat pipeline and the valve port node temperature and the inlet temperature of each process are collected continuously. The ambient temperature and waste heat flow rate are also recorded. A temperature prediction model is constructed based on past valve node temperatures, process inlet temperatures, ambient temperatures, and waste heat flow rates to obtain the predicted inlet temperatures of each process level at several future time points. The predicted inlet temperatures are compared with the temperature ranges of the corresponding processes to obtain the single-step fitness of each process level at each future time point. The overall fitness of the predicted inlet temperatures of each process level at several future time points is obtained by combining the changes in the single-step fitness of each process level at consecutive time points before each future time point. The distribution relationship between the predicted inlet temperature and the optimal temperature range of the corresponding process is analyzed to obtain the single-step efficiency impact and single-step lifetime impact of each process level at each future time. The impact and waste heat flow analysis within several future windows are then performed to obtain the comprehensive impact of the predicted inlet temperature of each process level in several future windows. Based on the overall fitness and the comprehensive impact, combined with the pipeline change relationship between each window and each process level, the comprehensive score of each process level in each future window is obtained. Based on the comprehensive evaluation of the processes at each level and in each window in the future, control the switching of pipelines and operate the staged membrane distillation water production process.

[0005] Optionally, the specific method for constructing the temperature prediction model includes: The valve inlet node temperature, ambient temperature, and waste heat flow rate of the past 7 days are input into the model, and the output is the inlet temperature of each process at the corresponding time. An LSTM network is constructed, and the valve inlet node temperature, inlet temperature of each process, ambient temperature, and waste heat flow rate of each node are divided into training set, validation set, and test set according to the proportion. The training stops when the MSE of the validation set is stable, and the model is saved.

[0006] Optionally, the specific method for obtaining the predicted inlet temperature of each process stage at several future time points includes: Input the current and past 10-minute valve inlet temperature, ambient temperature, and waste heat flow rate into the trained model to predict the inlet temperature of each process level every 10 seconds within the next 10 minutes, which will be used as the predicted inlet temperature of each process level at several future moments.

[0007] Optionally, the specific method for obtaining the single-step fitness of each process level at future time points includes: Define the temperature range of the three-level process. For the predicted inlet temperature of any level of the process at any future time, if the predicted inlet temperature is within the temperature range of the process, obtain the median and range of the temperature range of the process. The ratio of the absolute value of the difference between the predicted inlet temperature and the median to the range is used as the deviation of the process at that future time. The difference obtained by subtracting the deviation from 1 is used as the single-step fitness of the process at that future time. If the predicted inlet temperature is not within the temperature range of the process, the deviation of the predicted inlet temperature from the upper and lower limits of the temperature range is obtained in the future at that time. The difference between 0.5 and the deviation is compared with 0. The maximum value of the two is taken as the single-step fitness of the process in the future at that time.

[0008] Optionally, the specific method for obtaining the overall fitness of the predicted inlet temperature of each process level in several future windows includes: Starting from the current moment, a sliding window is constructed, dividing several future moments into several windows; for any window, the last moment in the window is taken as the switching moment of the window, the time distance between each moment in the window and the switching moment is obtained, the difference obtained by subtracting the time distance from the total duration of the window is taken as the time difference coefficient of each moment in the window, and the weight normalization is performed on the time difference coefficients of all moments, and the result is taken as the time weight of each moment in the window. The single-step fitness of the window for any level of process at each time point is weighted and summed using time weights. The result is used as the overall fitness of the window for the inlet predicted temperature of the process.

[0009] Optionally, the specific method for obtaining the single-step efficiency impact and single-step lifetime impact of each process stage at each future time point is as follows: Define the optimal temperature range for the three-stage process. For any future moment, if the predicted inlet temperature of any stage of the process is within the optimal temperature range of the process, the single-step efficiency impact and single-step lifetime impact of the process at that moment are both 0. If the predicted inlet temperature is less than the lower limit of the optimal temperature range, the single-step lifetime impact of the process at that moment is set to 0. The difference between the lower limit of the optimal temperature range and the predicted inlet temperature is obtained. The product of this difference and 0.008 is used as the single-step efficiency impact of the process at that moment. If the predicted inlet temperature is greater than the upper limit of the optimal temperature range, the single-step efficiency impact of the process at that moment is set to 0. The difference between the predicted inlet temperature and the upper limit of the optimal temperature range is obtained and divided by 5°C to obtain the ratio. The product of this ratio and 0.1 is used as the single-step lifetime impact of the process at that moment.

[0010] Optionally, the specific method for obtaining the comprehensive influence of the predicted inlet temperature of each process level in several future windows includes: For any time within any window, the sum of the single-step efficiency impact and the single-step lifetime impact at that time is taken as the single-step impact at that time. The single-step influence of each moment is weighted and summed by the time weight of each moment in the window, and the result is used as the overall influence of the window on the inlet predicted temperature of the process. The proportion of waste heat flow to total water volume within the window is statistically analyzed, and the product of this proportion and the overall influence is used as the comprehensive influence of this window on the predicted inlet temperature of the process.

[0011] Optionally, the comprehensive score of each stage of the process in the future window is obtained by the following method: For any window, the difference between the overall fitness of the window at any level of process and the comprehensive influence is taken as the fitness score of the window at that process. The pipe corresponding to the previous window adjacent to this window is taken as the current pipe of this window. If the pipe corresponding to this window in this process is different from the current pipe, the switching cost of this window in this process is set to 0.1; otherwise, it is set to 0. The pipe corresponding to the minimum adaptation score of each window in each process level is taken as the erroneous pipe of each window. If the pipes used in the two windows before this window are both erroneous pipes, the continuous mismatch penalty factor of this window is set to 0.2; otherwise, it is set to 0. The window's overall score for the process is obtained by subtracting the switching cost from its adaptation score for that process, and then subtracting the continuous mismatch penalty factor for that window.

[0012] Optionally, the control pipeline switching includes the following specific methods: For any window, the pipeline corresponding to the process with the highest comprehensive score among all processes in that window is taken as the target pipeline for that window. If the difference between the maximum value and the comprehensive score of the process corresponding to the current pipeline in the window is less than or equal to the preset switching threshold, the current pipeline will be used as the target pipeline of the window.

[0013] The present invention also proposes a waste heat water production system based on membrane distillation, the system comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of the above method.

[0014] The beneficial effects of this invention are as follows: This invention obtains the temperature change process of waste hot water from before the valve to the membrane distillation process end through end-to-end temperature monitoring, providing a real and comprehensive original data source for subsequent prediction and evaluation, avoiding the one-sidedness of single-point monitoring; the waste heat will naturally cool down during pipeline transmission, and the temperature fluctuation before the valve directly affects the subsequent pipeline distribution, and multi-point coverage can capture the spatial and temporal characteristics of temperature change; according to the waste heat distributed in the pipeline, the optimal membrane distillation process is operated accordingly, making full use of waste heat in different temperature ranges to achieve efficient water production. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a schematic diagram of a waste heat water production method based on membrane distillation, provided as an embodiment of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Please see Figure 1 The diagram illustrates a flow chart of a waste heat water production method based on membrane distillation according to an embodiment of the present invention. The method includes the following steps: Step S001: Arrange several nodes in the main waste heat pipeline and collect the valve port node temperature at continuous intervals, as well as the inlet temperature of each process level, and record the ambient temperature and waste heat flow rate.

[0019] It should be noted that by monitoring the temperature of the waste water throughout the entire process, the temperature change process from the valve port to the membrane distillation process end can be obtained, providing a real and comprehensive raw data source for subsequent prediction and evaluation, and avoiding the one-sidedness of monitoring at a single point. The waste heat will naturally cool down during pipeline transmission, and the temperature fluctuation before the valve port directly affects the subsequent pipeline distribution. Multi-point coverage can capture the spatial and temporal characteristics of temperature change.

[0020] Specifically, on the main waste heat pipeline, one PT100 high-precision temperature sensor (error ±0.2℃) is installed at distances of 0m (before the valve), 5m, 10m, and 15m from the pipeline switching valve. Simultaneously, one sensor is installed at each process inlet of the DCMD, VMD, and AGMD stages. The sampling frequency is set to 1 time / second, and all sensor data is synchronously uploaded to the data processing terminal via an RS485 bus. For the first Each monitoring point is located at Temperature at any given moment, among which Corresponding to nodes 0m, 5m, 10m and 15m before the valve port Corresponding to the process entry points of DCMD, VMD, and AGMD.

[0021] Furthermore, a 5-second sliding window mean filtering method is employed: for the time-series data at each point, the mean is calculated using 5 consecutive sampling points as a window. If the deviation between the current sampling point and the window mean exceeds ±5℃, it is identified as outlier data and removed. Missing data is filled in using linear interpolation (i.e., filling in missing values ​​based on the temperature difference between two consecutive valid data points). Outliers with temperatures exceeding 30-100℃ are removed (the residual heat temperature range is typically 40-90℃; exceeding this range is considered a sensor malfunction). The temperature data is then normalized to... The interval is convenient for model processing; it enables the acquisition of valve node temperature at several times and inlet temperature of each process level; at the same time, the ambient temperature at each time is acquired and normalized, and the waste heat flow rate at the corresponding time is recorded.

[0022] Step S002: Based on past valve node temperatures, process inlet temperatures, ambient temperatures, and waste heat flow, construct a temperature prediction model to obtain the predicted inlet temperatures of each process level at several future time points; compare the predicted inlet temperatures with the corresponding process temperature ranges to obtain the single-step fitness of each process level at each future time point; combine the single-step fitness changes of each process level at consecutive time points before each future time point to obtain the overall fitness of the predicted inlet temperatures of each process level at several future windows.

[0023] It should be noted that by acquiring temperature data from different valve nodes, the time-series memory capability of LSTM can be used to predict the temperature change trend of waste heat at the valve in the next 30 seconds, thereby formulating pipeline switching strategies in advance and avoiding decision lag. The waste heat temperature is affected by ambient temperature, waste heat flow rate, and pipeline insulation performance, and multiple feature inputs can improve prediction accuracy. The gating structure of LSTM can memorize long-term temperature change patterns, avoiding the gradient vanishing problem of traditional time-series models.

[0024] Preferably, in one embodiment of the present invention, a temperature prediction model is constructed based on past valve inlet temperatures, process inlet temperatures, ambient temperatures, and waste heat flow rates to obtain the predicted inlet temperatures of each process stage at several future times. The specific method includes: The valve inlet node temperature, ambient temperature, and waste heat flow rate over the past 7 days were input into the model, and the output was the inlet temperature of each process stage at the corresponding time. An LSTM network was constructed with 2 hidden layers (64 neurons per layer), dropout=0.2 to prevent overfitting, and the Adam optimizer was used with the mean squared error (MSE) as the loss function. The valve inlet node temperature, inlet temperature of each process stage, ambient temperature, and waste heat flow rate were divided into training, validation, and test sets in a ratio of 7:2:1. Training was stopped when the MSE of the validation set was stably below 0.001, and the model was saved.

[0025] Furthermore, the valve inlet node temperature, ambient temperature, and waste heat flow rate of the current and past 10 minutes are input into the trained model to predict the inlet temperature of each process level every 10 seconds within the next 10 minutes, which will be used as the predicted inlet temperature of each process level at several future moments.

[0026] It should be noted that the time for waste heat to be transferred from the valve port to the process inlet is about 2-3 minutes. Predicting 10 minutes in advance can allow sufficient preparation time for pipeline switching and avoid delayed control.

[0027] It should be further noted that the three-stage membrane process has overlapping temperature ranges (such as 60°C, which falls under both DCMD and VMD). The temperature range of the overlapping range needs to be reasonably matched. When the temperature deviates from the process range, the efficiency or lifespan will be affected. Therefore, a lower matching degree should be given instead of complete exclusion.

[0028] Preferably, in one embodiment of the present invention, comparing the predicted inlet temperature with the corresponding process temperature range to obtain the single-step fitness of each process level at future time points includes the following specific method: The temperature ranges for the three-level process are defined as DCMD (60-100℃), VMD (40-80℃), and AGMD (30-60℃) in this embodiment. For the predicted inlet temperature of any level process at any future time, if the predicted inlet temperature is within the temperature range of the process, the median (mean of the upper and lower limits of the range) and the range (difference between the upper and lower limits of the range) of the temperature range of the process are obtained. The ratio of the absolute value of the difference between the predicted inlet temperature and the median to the range is used as the deviation of the process at that future time. The difference obtained by subtracting the deviation from 1 is used as the single-step fitness of the process at that future time.

[0029] Furthermore, if the predicted inlet temperature is greater than the upper limit of the temperature range of the process, the ratio of the difference between the predicted inlet temperature and the upper limit (in °C) to 20 °C is used as the deviation of the process at that future moment. The difference between 0.5 and the deviation is compared with 0, and the maximum value of the two is used as the single-step fitness of the process at that future moment. Conversely, if the predicted inlet temperature is less than the lower limit of the temperature range of the process, the ratio of the difference between the lower limit and the predicted inlet temperature (in °C) to 20 °C is used as the deviation of the process at that future moment. The difference between 0.5 and the deviation is compared with 0, and the maximum value of the two is used as the single-step fitness of the process at that future moment.

[0030] It should be noted that the closer the predicted inlet temperature is to the midpoint (median) within the temperature range, the higher the fitness (maximum 1). When the predicted inlet temperature is within the temperature range, the fitness ranges from a maximum of 1 to a minimum of 0.5. When the inlet temperature is outside the temperature range, the fitness decreases by 0.05 for every 1°C deviation from the outside temperature, down to a minimum of 0, and is truncated to zero using a maximum value function.

[0031] It should be further noted that since the predicted temperature is time-series data, it needs to be processed in conjunction with the time-series temperature data. When short-term abrupt changes occur in the temperature data, they will significantly affect the single-step fit mentioned above and need to be processed. Short-term abrupt changes (such as temperature jumps at 1-2 sampling points) only affect the moment and do not represent the overall characteristics of the residual heat during that period. The abrupt changes can be smoothed out by using a sliding window weighted average, while highlighting the dominant role of recent temperature.

[0032] Preferably, in one embodiment of the present invention, the overall fitness of the inlet predicted temperature of each stage of the process in the future several windows is obtained by combining the single-step fitness changes of each stage of the process in consecutive time periods before each future time period. The specific method includes: Starting from the current moment, a sliding window is constructed. In this embodiment, the sliding window is described in 3-minute intervals. A non-overlapping division is used to divide several future moments into several windows, where each future moment is 10 seconds long. The first future moment is the first moment in the first window. Each window includes 18 moments in total. For any window, the last moment in the window is taken as the switching moment of the window. The time distance between each moment in the window and the switching moment is obtained. The difference between the total window duration and the time distance is taken as the time difference coefficient of each moment in the window. The time difference coefficients of all moments are weighted and normalized, and the result is taken as the time weight of each moment in the window. The single-step fitness of each moment in the window for any level of process is weighted and summed using the time weights. The result is taken as the overall fitness of the window for the inlet predicted temperature of the process.

[0033] For all time difference coefficients mentioned above, weight normalization is performed. In a specific embodiment of the present invention, the specific method of weight normalization is as follows: obtain the cumulative value of all time difference coefficients within the window as the total time difference coefficient of the window, and use the ratio of the time difference coefficient of any time within the window to the total time difference coefficient as the time weight of the corresponding time.

[0034] Thus, the overall adaptability of the predicted inlet temperature for each stage of the process in several future windows is obtained.

[0035] Step S003: Analyze the distribution relationship between the predicted inlet temperature and the optimal temperature range of the corresponding process, obtain the single-step efficiency impact and single-step lifetime impact of each process level at each future time, and perform impact and waste heat flow analysis within several future windows to obtain the comprehensive impact of the predicted inlet temperature of each process level in several future windows; based on the overall fitness and the comprehensive impact, combined with the pipeline change relationship between each window and each process level, obtain the comprehensive score of each process level in each future window.

[0036] It should be noted that quantifying the impact of temperature deviation from the optimal process range on membrane lifespan loss and water production efficiency reduction, combined with the proportion of waste heat flow to assess the actual degree of loss, provides a basis for loss cost in control strategies. When the temperature is below the optimal process range, the mass transfer coefficient of membrane distillation decreases, and the water production efficiency decreases linearly (experiments show that efficiency decreases by about 4% for every 5°C drop in temperature). When the temperature is above the optimal process range, the polymer chains of hydrophobic membranes age faster, and membrane lifespan shortens exponentially (experiments show that membrane lifespan shortens by about 10% for every 5°C increase in temperature).

[0037] Preferably, in one embodiment of the present invention, the distribution relationship between the predicted inlet temperature and the optimal temperature range of the corresponding process is analyzed to obtain the single-step efficiency impact and single-step lifetime impact of each process level at future time points, and the impact and waste heat flow analysis are performed within several future windows to obtain the comprehensive impact of the predicted inlet temperature of each process level in several future windows. The specific method includes: The optimal temperature ranges for the three-stage process are defined in this embodiment as DCMD (65-90℃), VMD (50-70℃), and AGMD (35-55℃). For any future moment, if the predicted inlet temperature of any stage of the process is within the optimal temperature range of the process, the single-step efficiency impact and single-step lifetime impact of the process at that moment are both 0. If the predicted inlet temperature is less than the lower limit of the optimal temperature range, the single-step lifetime impact of the process at that moment is set to 0. The difference between the lower limit of the optimal temperature range and the predicted inlet temperature is obtained, and the product of this difference and 0.008 is used as the single-step efficiency impact of the process at that moment.

[0038] Furthermore, if the predicted inlet temperature is greater than the upper limit of the optimal temperature range, the single-step efficiency impact of the process at that moment is set to 0. The difference between the predicted inlet temperature and the upper limit of the optimal temperature range is obtained, and the difference is divided by 5°C to obtain the ratio. The product of this ratio and 0.1 is used as the single-step lifetime impact of the process at that moment.

[0039] It should be noted that there is no efficiency loss within the optimal range. When the temperature drops below the optimal lower limit, the efficiency loss is approximately 0.8% for every 1°C decrease, with a maximum loss not exceeding 80% (in reality, the waste heat temperature will not be lower than 10°C). There is no lifespan loss within the optimal range. When the temperature rises above the optimal upper limit, the membrane lifespan loss is 10% for every 5°C increase, to avoid rapid membrane aging caused by high temperatures. The larger the waste heat flow rate, the more significant the loss caused by temperature fluctuations, and the actual impact needs to be assessed in conjunction with the water volume ratio.

[0040] Furthermore, for any given moment within any window, the sum of the single-step efficiency impact and the single-step lifetime impact at that moment is taken as the single-step impact at that moment. The single-step impact at each moment within the window is weighted and summed using the time weights of each moment, and the result is taken as the overall impact of the window on the inlet predicted temperature of the process. The proportion of waste heat flow to total water volume within the window is calculated (monitored by a flow sensor; the total water volume is defined as the theoretical integral of the maximum rated flow of the main pipeline of the system within the 3-minute sliding window period). The product of the proportion and the overall impact is taken as the comprehensive impact of the window on the inlet predicted temperature of the process.

[0041] It should be further explained that, by combining adaptability, impact and switching costs, a multi-objective decision-making algorithm is used to select the optimal pipeline, balancing process adaptability, membrane loss and switching frequency, and maximizing the utilization of waste heat.

[0042] Preferably, in one embodiment of the present invention, based on the overall fitness and the comprehensive influence, and combined with the pipeline change relationship between each window and each level of process, the comprehensive score of each window and each level of process in the future is obtained, including the following specific method: For any window, the difference between the overall fitness score and the comprehensive impact score corresponding to the window in any process level is taken as the fitness score of the window in that process. The pipe corresponding to the window adjacent to the previous window is taken as the current pipe of the window (the current pipe corresponding to the first window is the pipe at the current moment). If the pipe corresponding to the window in that process is different from the current pipe, the switching cost of the window in that process is set to 0.1. If they are the same, it is set to 0. The pipe corresponding to the minimum fitness score of each window in each process level is taken as the erroneous pipe of each window. If the pipes used in the two windows before the window are both erroneous pipes, the continuous mismatch penalty factor of the window is set to 0.2. Otherwise, it is set to 0.

[0043] Furthermore, the window's adaptation score for the process is subtracted by the switching cost, and then by the continuous mismatch penalty factor, to obtain the overall score for the window in the process.

[0044] It should be noted that the higher the compatibility and the lower the impact, the higher the score; switching once increases valve wear, sets a fixed cost, and avoids long-term mismatch leading to serious losses.

[0045] At this point, a comprehensive score for each window and each level of process is obtained.

[0046] Step S004: Based on the comprehensive score of each window and each level of the process in the future, control the pipeline switching and run the staged membrane distillation water production process.

[0047] Specifically, for any window, the pipeline corresponding to the process with the maximum value in the comprehensive score of each process level in the window is taken as the target pipeline of the window; a preset switching threshold is set, and in this embodiment, the switching threshold is described as 0.05; if the difference between the maximum value and the comprehensive score of the process corresponding to the current pipeline in the window (the maximum value minus the comprehensive score corresponding to the current pipeline) is less than or equal to the switching threshold, the current pipeline is taken as the target pipeline of the window, that is, process switching is reduced.

[0048] Furthermore, a switching command is sent 30 seconds in advance to ensure that the switching is completed when the waste heat reaches the process inlet, avoiding the entry of waste heat into the process during mismatched periods; the pipeline switching is executed by an electric regulating valve, and the control signal is issued by the PLC. Based on the decision result at the end of the window, the switching command is sent 30 seconds in advance, and the regulating valve response time is <10 seconds; the system sets the minimum steady-state operation cycle to 30 minutes. If the current process pipeline has not been running continuously for 30 minutes, the switching command is frozen, the difference in comprehensive score is ignored, and the current pipeline operation state is forcibly maintained.

[0049] Furthermore, based on the waste heat allocated through the pipeline, the optimal membrane distillation process is operated to fully utilize the waste heat in different temperature ranges and achieve efficient water production. The specific process flow is as follows: DCMD process operation (high temperature range: 60-100℃): DCMD has the highest mass transfer efficiency in the high-temperature range, and direct contact condensation can quickly recover water vapor, making it suitable for high-grade waste heat.

[0050] 1) Waste heat hot water enters the hot side of the DCMD module, with a flow rate controlled at 1.5-2.0 m³ / h; 2) At room temperature clean water is introduced into the cold side at the same flow rate as the hot side, and the membrane module operates at atmospheric pressure; 3) The remaining wastewater on the hot side is returned to the waste heat system or discharged in compliance with standards, and the fresh water obtained from the condensation on the cold side is collected into the water storage tank.

[0051] VMD process operation (intermediate temperature range: 40-80℃): The vacuum environment reduces the partial pressure of water vapor, improves the mass transfer efficiency in the medium temperature range, avoids excessively low efficiency at low temperatures, and is suitable for medium-grade waste heat.

[0052] 1) Waste heat hot water enters the thermal side of the VMD module, with a flow rate controlled at 1.0-1.5 m³ / h; 2) A vacuum pump is connected to the membrane permeation side to maintain a vacuum of 0.08-0.09 MPa, and water vapor is condensed into fresh water in the condenser; 3) The vacuum pump frequency is automatically adjusted according to the vacuum level to ensure vacuum stability.

[0053] AGMD process operation (low temperature range: 30-60℃): The air gap structure reduces heat conduction loss, improves water production efficiency in the low-temperature section, and is suitable for low-grade waste heat.

[0054] 1) Waste heat hot water enters the hot side of the AGMD component, with a flow rate controlled at 0.8-1.2 m³ / h; 2) There is a 1-2mm air gap between the membrane and the condenser plate. The temperature of the condenser plate is maintained at 15-20℃. Water vapor passes through the membrane pores and condenses into fresh water on the condenser plate. 3) The temperature of the condenser plate is automatically regulated by the cold water circulation system.

[0055] Furthermore, bypass overflow valves are installed at the process inlets of DCMD, VMD, and AGMD. When the actual waste heat flow in the pipeline exceeds the upper limit of the flow set for a single process, the overflow valve automatically opens, and the excess flow is discharged into the low-temperature standby water storage tank. When the flow is lower than the set lower limit, the frequency of the process-side pump is reduced to match the current flow.

[0056] This concludes the embodiment.

[0057] Another embodiment of the present invention provides a waste heat water production system based on membrane distillation. The system includes a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the above-described method steps S001 to S004.

[0058] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A waste heat water production method based on membrane distillation, characterized in that, The method includes the following steps: Several nodes are arranged in the main waste heat pipeline and the valve port node temperature and the inlet temperature of each process are collected continuously. The ambient temperature and waste heat flow rate are also recorded. A temperature prediction model is constructed based on past valve node temperatures, process inlet temperatures, ambient temperatures, and waste heat flow rates to obtain the predicted inlet temperatures of each process level at several future time points. The predicted inlet temperatures are compared with the temperature ranges of the corresponding processes to obtain the single-step fitness of each process level at each future time point. The overall fitness of the predicted inlet temperatures of each process level at several future time points is obtained by combining the changes in the single-step fitness of each process level at consecutive time points before each future time point. The distribution relationship between the predicted inlet temperature and the optimal temperature range of the corresponding process is analyzed to obtain the single-step efficiency impact and single-step lifetime impact of each process level at each future time. The impact and waste heat flow analysis within several future windows are then performed to obtain the comprehensive impact of the predicted inlet temperature of each process level in several future windows. Based on the overall fitness and the comprehensive impact, combined with the pipeline change relationship between each window and each process level, the comprehensive score of each process level in each future window is obtained. Based on the comprehensive evaluation of the processes at each window and level in the future, control the switching of pipelines and operate the staged membrane distillation water production process.

2. The waste heat water production method based on membrane distillation according to claim 1, characterized in that, The specific methods for constructing the temperature prediction model are as follows: The valve inlet node temperature, ambient temperature, and waste heat flow rate of the past 7 days are input into the model, and the output is the inlet temperature of each process at the corresponding time. An LSTM network is constructed, and the valve inlet node temperature, inlet temperature of each process, ambient temperature, and waste heat flow rate of each node are divided into training set, validation set, and test set according to the proportion. The training stops when the MSE of the validation set is stable, the model is saved, and it is used as a temperature prediction model.

3. The waste heat water production method based on membrane distillation according to claim 1, characterized in that, The specific method for obtaining the predicted inlet temperature of each process stage at a future time is as follows: The current and past valve node temperatures, ambient temperatures, and waste heat flow rates are input into the trained model to predict the inlet temperatures of each process level at each sampling interval in the future, which will be used as the predicted inlet temperatures of each process level at several future times.

4. The waste heat water production method based on membrane distillation according to claim 1, characterized in that, The specific methods for obtaining the single-step fitness of each level of the process at future time points include: Define the temperature range of the three-level process. For the predicted inlet temperature of any level of the process at any future time, if the predicted inlet temperature is within the temperature range of the process, obtain the median and range of the temperature range of the process. The ratio of the absolute value of the difference between the predicted inlet temperature and the median to the range is used as the deviation of the process at that future time. The difference obtained by subtracting the deviation from 1 is used as the single-step fitness of the process at that future time. If the predicted inlet temperature is not within the temperature range of the process, the deviation of the predicted inlet temperature from the upper and lower limits of the temperature range is obtained in the future at that time. The difference between 0.5 and the deviation is compared with 0. The maximum value of the two is taken as the single-step fitness of the process in the future at that time.

5. A waste heat water production method based on membrane distillation according to claim 1, characterized in that, The specific method for obtaining the overall adaptability of the predicted inlet temperature of each process level in several future windows includes: Starting from the current moment, a sliding window is constructed, dividing several future moments into several windows; for any window, the last moment in the window is taken as the switching moment of the window, the time distance between each moment in the window and the switching moment is obtained, the difference obtained by subtracting the time distance from the total duration of the window is taken as the time difference coefficient of each moment in the window, and the weight normalization is performed on the time difference coefficients of all moments, and the result is taken as the time weight of each moment in the window. The single-step fitness of the window for any level of process at each time point is weighted and summed using time weights. The result is used as the overall fitness of the window for the inlet predicted temperature of the process.

6. The waste heat water production method based on membrane distillation according to claim 1, characterized in that, The specific method for obtaining the single-step efficiency impact and single-step lifetime impact of each process level at each future time point is as follows: Define the optimal temperature range for the three-stage process. For any future moment, if the predicted inlet temperature of any stage of the process is within the optimal temperature range of the process, the single-step efficiency impact and single-step lifetime impact of the process at that moment are both 0. If the predicted inlet temperature is less than the lower limit of the optimal temperature range, the single-step lifetime impact of the process at that moment is set to 0. The difference between the lower limit of the optimal temperature range and the predicted inlet temperature is obtained. The product of this difference and 0.008 is used as the single-step efficiency impact of the process at that moment. If the predicted inlet temperature is greater than the upper limit of the optimal temperature range, the single-step efficiency impact of the process at that moment is set to 0. The difference between the predicted inlet temperature and the upper limit of the optimal temperature range is obtained and divided by 5°C to obtain the ratio. The product of this ratio and 0.1 is used as the single-step lifetime impact of the process at that moment.

7. The waste heat water production method based on membrane distillation according to claim 1, characterized in that, The method for obtaining the comprehensive influence of the predicted inlet temperature of each process level in several future windows includes: For any time within any window, the sum of the single-step efficiency impact and the single-step lifetime impact at that time is taken as the single-step impact at that time. The single-step influence of each moment is weighted and summed by the time weight of each moment in the window, and the result is used as the overall influence of the window on the inlet predicted temperature of the process. The proportion of waste heat flow to total water volume within the window is statistically analyzed, and the product of this proportion and the overall influence is used as the comprehensive influence of this window on the predicted inlet temperature of the process.

8. A waste heat water production method based on membrane distillation according to claim 1, characterized in that, The comprehensive score of each window and each level of process in the future is obtained by the following method: For any window, the difference between the overall fitness of the window at any level of process and the comprehensive influence is taken as the fitness score of the window at that process. The pipe corresponding to the previous window adjacent to this window is taken as the current pipe of this window. If the pipe corresponding to this window in this process is different from the current pipe, the switching cost of this window in this process is set to 0.1; otherwise, it is set to 0. The pipe corresponding to the minimum adaptation score of each window in each process level is taken as the erroneous pipe of each window. If the pipes used in the two windows before this window are both erroneous pipes, the continuous mismatch penalty factor of this window is set to 0.2; otherwise, it is set to 0. The window's overall score for the process is obtained by subtracting the switching cost from its adaptation score for that process, and then subtracting the continuous mismatch penalty factor for that window.

9. A waste heat water production method based on membrane distillation according to claim 8, characterized in that, The specific methods for switching the control pipeline are as follows: For any window, the pipeline corresponding to the process with the highest comprehensive score among all processes in that window is taken as the target pipeline for that window. If the difference between the maximum value and the comprehensive score of the process corresponding to the current pipeline in the window is less than or equal to the preset switching threshold, the current pipeline will be used as the target pipeline of the window.

10. A waste heat water production system based on membrane distillation, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the waste heat water production method based on membrane distillation as described in any one of claims 1-9.

Images