CN-122015062-A - Factory production steam and heating ventilation coupling control method based on neural network

CN122015062ACN 122015062 ACN122015062 ACN 122015062ACN-122015062-A

Abstract

The application relates to the technical field of industrial control, and discloses a factory production steam and heating ventilation coupling control method based on a neural network, which comprises the steps of collecting source side, charge side and state side parameters of a multi-energy coupling cascade utilization system and constructing a time sequence feature vector; the method comprises the steps of firstly, obtaining a physical model of stratum seepage, then utilizing a hard parameter sharing neural network model to decouple and output steam flow, heating and ventilation heat load and stratum recharging pressure response predicted values, calculating the maximum recharging flow in real time as a safety boundary according to recharging pressure response and the stratum seepage physical model, further constructing a discrete state space model, mapping the predicted value of the requirement to a state variable reference track, finally under the constraint of the maximum recharging flow, driving mechanisms such as a submersible pump, a peak shaving device and the like to act through a control parameter sequence which aims at minimizing operation cost and tracking error through the discrete state space model. The application effectively considers the geothermal exploitation safety and the cascade cooperative utilization efficiency of different grade heat energy.

Inventors

SU JINGQIN
AN DI
YANG BO
CHEN WENQIANG
ZHANG LEI
XU QIANGQIANG

Assignees

中煤科工集团信息技术有限公司

Dates

Publication Date: 20260512
Application Date: 20251231

Claims (10)

1. The factory production steam and heating ventilation coupling control method based on the neural network is realized based on a multi-energy coupling cascade utilization system, and the multi-energy coupling cascade utilization system comprises a deep geothermal energy supply loop, an auxiliary steam production loop and a heating ventilation air conditioning loop which are physically connected, and is characterized by comprising the following steps: Acquiring source side parameters, load side parameters and state side parameters of the multi-energy coupling cascade utilization system, performing time alignment and standardization processing on the acquired data, and constructing a time sequence feature vector reflecting the current running state of the multi-energy coupling cascade utilization system; Inputting the time sequence feature vector into a pre-trained neural network model, and outputting a steam flow demand predicted value, a heating ventilation heat load demand predicted value and a stratum recharging pressure response predicted value in a future preset time period; According to the stratum recharging pressure response predicted value, combining a stratum seepage physical model constructed based on Darcy's law and a non-Darcy flow correction theory, and calculating the maximum recharging flow of the deep geothermal system at the current moment as a safety boundary; Determining state variables used for representing the running states of the auxiliary steam-producing loop and the heating ventilation air conditioning loop, and establishing a discrete state space model describing the dynamic change relation of the state variables; Taking the maximum recharging flow as a safety constraint condition of a dynamic flow boundary at a source side, and solving an optimal control parameter sequence of each executing mechanism in a future limited domain by taking the minimum system running cost as an objective function by utilizing the discrete state space model on the premise of meeting the rigidity requirement of a flow demand predicted value of factory production steam; And decomposing the optimal control parameter sequence into control instructions aiming at the bottom layer controller, and issuing the control instructions to an executing mechanism to drive physical equipment to act so as to realize coupling control.
2. The factory production steam and heating ventilation coupling control method based on the neural network according to claim 1, wherein the deep geothermal energy supply loop comprises a deep mining well (1), a submerged pump (4) is arranged in the deep mining well (1), an outlet of the submerged pump (4) is connected with an inlet on one side of a primary plate heat exchanger (2) through a conveying pipe, an outlet on one side of the primary plate heat exchanger (2) is connected with a deep recharging well (3) through a pipeline, the auxiliary steam production loop comprises a steam generator (5) and a fuel gas peak shaving device (6), a secondary side outlet of the primary plate heat exchanger (2) is connected to a water inlet of the steam generator (5), the heating ventilation air conditioning loop comprises a shallow geothermal heat pump unit (7), a tail end air conditioning device (8) and a shallow geothermal heat exchange pipe group (9), and a primary side outlet pipeline of the primary plate heat exchanger (2) is connected with a water supply side of the tail end air conditioning device (8) through a three-way regulating valve (10).
3. The method for controlling the coupling of factory production steam and heating ventilation according to claim 2, wherein the source side parameters comprise wellhead fluid temperature of a deep mining well (1), current mining flow of the deep mining well (1), wellhead pressure of a deep recharging well (3) and liquid level of the deep recharging well (3), the load side parameters comprise factory production schedule time sequence vectors, real-time steam pipe network pressure, real-time steam pipe network flow, indoor average temperature, outdoor dry bulb temperature and solar radiation intensity, and the state side parameters comprise opening feedback of a three-way regulating valve (10), operation frequency feedback of a submersible pump (4), real-time operation power of a gas peak regulating device (6) and start-stop state of a shallow geothermal heat pump unit (7).
4. The plant production steam and heating ventilation coupling control method based on the neural network according to claim 1, wherein the neural network model is constructed by adopting a hard parameter sharing mechanism, and comprises a shared feature extraction layer and a task specific decoding layer; The shared feature extraction layer is composed of a plurality of layers of long-term and short-term memory network units and is used for extracting shared hidden space feature vectors in the time sequence feature vectors; The task specific decoding layer comprises a steam rigidity demand prediction branch, a heating ventilation flexible demand prediction branch and a recharging pressure response prediction branch, wherein each branch receives the shared hidden space feature vector respectively and correspondingly outputs the steam flow demand prediction value, the heating ventilation heat load demand prediction value and the stratum recharging pressure response prediction value.
5. The neural network-based plant-produced steam and heating ventilation coupling control method of claim 2, wherein determining the state variables and constructing the discrete state space model comprises: selecting the internal pressure of the steam generator (5) as one of the state variables, and constructing a steam generation stage dynamic equation, wherein the change rate of the internal pressure of the steam generator (5) is related to real-time geothermal fluid flow and actual plant production steam flow demand values; selecting the warm water supply temperature of the shallow geothermal heat pump unit (7) as the second state variable, and constructing a warm ventilation heating stage dynamic equation, wherein the change rate of the warm water supply temperature is related to the real-time flow of the geothermal fluid, the temperature of heating backwater and the flow of heating circulating water; And the steam generation stage dynamic equation and the heating stage dynamic equation are combined and discretized to obtain the discrete state space model comprising a state transition matrix, an input control matrix and a disturbance response matrix.
6. The method for controlling the coupling of steam and heating ventilation in a plant based on a neural network according to claim 1, wherein the process of the maximum recharging flow comprises the following steps: Based on Darcy's law and non-Darcy flow correction theory, establishing a linearization engineering calculation model for describing the relation between the flow pressure of the deep recharging well (3), the static pressure of the stratum, the dynamic recharging impedance coefficient and the real-time flow of geothermal fluid; a recursive least square method with forgetting factors is adopted, and the gain is calculated and the estimated value of the dynamic recharging impedance coefficient is updated on line by utilizing the real-time collected effective differential pressure observed value and the input regression scalar; Setting a maximum safe pressure threshold and a safety margin coefficient of the deep recharging well (3), and obtaining a source side maximum recharging flow boundary at the current moment by calculating the difference between the maximum safe pressure threshold of the deep recharging well (3) and the stratum static pressure and combining the safety margin coefficient to perform reverse operation by utilizing the updated dynamic recharging impedance coefficient.
7. The neural network-based plant production steam and heating ventilation coupling control method according to claim 1, wherein the constraint conditions include: controlling the amplitude constraint of the quantity, directly limiting the exploitation flow of the deep exploitation well (1) by using the boundary of the maximum recharging flow, and ensuring that the exploitation action is always in the geological safety boundary; The control quantity increment constraint is used for setting the maximum descending rate and the maximum ascending rate allowed by the executing mechanism according to the acceleration and deceleration time parameter of the frequency converter; Outputting soft constraint, setting lower limit and upper limit of state variable allowed by the process, and converting hard constraint into soft constraint by combining non-negative relaxation factor.
8. The neural network-based plant production steam and heating ventilation coupling control method according to claim 1, wherein the system running cost minimization objective function consists of a multi-objective weighting term, and specifically comprises: Outputting a tracking deviation term for calculating a weighted square sum of Euclidean distances between the system output vector and the reference trajectory vector, wherein the weight element value of the corresponding pressure is set to be greater than the weight element value of the corresponding temperature so as to establish a strategy of giving priority to steam supply stability; A control quantity increment term for calculating a weighted square sum of the control increment vectors to limit the action amplitude of the actuator; The soft constraint penalty term consisting of the relaxation factors allows the softening of the output constraint under extreme conditions by introducing a relaxation penalty weight to ensure that the optimization problem is solvable.
9. The neural network-based plant-produced steam and heating ventilation coupling control method of claim 2, further comprising automatically adjusting weight coefficients in the objective function according to an operating mode: when the multi-energy coupling cascade utilization system is in a heating and steam cooperative mode, setting a weight coefficient corresponding to the heating and water supply temperature in the objective function to be a non-zero positive value, so that a controller can track and optimize the internal pressure of the steam generator (5) and the heating and water supply temperature simultaneously; when the multi-energy coupling cascade utilization system is in a pure industrial steam mode, the weight coefficient corresponding to the heating and water supply temperature in the objective function is set to be zero, so that the controller only carries out tracking optimization on the internal pressure of the steam generator (5), and ignores the fluctuation of the heating and water supply temperature.
10. A neural network-based plant production steam and heating ventilation coupling control system, characterized in that the method is applied to the neural network-based plant production steam and heating ventilation coupling control method according to any one of claims 1-9, and comprises the following steps: The data acquisition processing module (101) is connected to the factory distributed control system and the manufacturing execution system, acquires source side, charge side and state side data, and performs cleaning, alignment and vectorization processing; The multitask prediction module (102) is connected with the data acquisition and processing module (101), and internally operates a neural network model with shared hard parameters and is used for outputting predicted values of steam flow demand, heating ventilation heat load demand and stratum recharging pressure response; The dynamic boundary calculation module (103) is connected to the multi-task prediction module (102) and is used for calculating the maximum recharging flow according to the stratum recharging pressure response predicted value and the stratum seepage physical model; The collaborative optimization module (104) is connected to the dynamic boundary calculation module (103) and the multi-task prediction module (102) and is used for solving an optimal control parameter sequence with minimized system running cost based on a discrete state space model under the constraint of a safety boundary and production requirements; The execution control module (105) is connected to the collaborative optimization module (104) and is used for receiving the optimal control parameter sequence and converting the optimal control parameter sequence into control instructions for driving the submersible pump (4), the three-way regulating valve (10), the gas peak regulating device (6) and the shallow geothermal heat pump unit (7) to act.

Description

Factory production steam and heating ventilation coupling control method based on neural network Technical Field The invention relates to the technical field of industrial control, in particular to a factory production steam and heating ventilation coupling control method based on a neural network. Background The deep geothermal energy is used as a clean and stable renewable energy source, and has wide application prospect in industrial steam supply and cascade utilization of building heating. Typical geothermal cascade utilization systems generally utilize high temperature geothermal fluids to produce industrial steam first, then utilize heat exchanged tail water to perform building heating, and finally recharge to underground reservoirs. However, such systems involve coupling of multiple physical domains such as deep geological environments, surface thermal equipment, and end user loads, and the operating mechanisms are complex, which presents a number of challenges for practical control. In the existing control method of the multi-energy coupling system, a mode of independent modeling is usually adopted for prediction of a source side and a load side. This approach often ignores the strong nonlinear coupling that exists between geothermal exploitation, industrial production scheduling, and building thermal inertia. For example, industrial steam demand is typically presented as a rigid, rapidly changing pulsed load, while building heating and ventilation loads have greater flexibility and hysteresis characteristics. The traditional method is difficult to effectively process the characteristics of the two different time scales in a unified model frame, so that the prediction result cannot accurately reflect the real dynamic response of the system under multiple working conditions, and the prospective and the accuracy of a control strategy are further affected. Furthermore, existing control strategies have significant shortcomings in geothermal resource exploitation safety. Conventional methods often employ fixed thresholds determined at the design stage to limit production flow or recharge pressure. However, the physical properties (such as permeability, skin factor) of deep geothermal reservoirs are not constant, but rather dynamically drift with the temperature change of the recharge fluid, suspended matter blockage, or chemical precipitation. If the control is performed only according to static parameters, geological safety risks such as wellhead overflow or bottom hole pressure exceeding rock fracture pressure and the like are very easy to occur when recharging impedance is increased, otherwise, too conservative setting can limit the capacity exertion of the system, and the energy utilization rate is reduced. In the cooperative optimization control level, the current control system focuses on meeting the set value tracking of a single loop, and lacks global overall planning of the whole cascade utilization chain. The existing control logic is difficult to dynamically adjust the control weight according to different operation scenes, and the stability requirement of industrial steam supply and the energy saving and consumption reduction requirement of a heating ventilation air conditioning system cannot be effectively balanced. Particularly, when the heat source is insufficient or the load fluctuation is large, an effective priority scheduling mechanism is lacked, so that the key industrial production cannot be preferentially ensured by high-grade heat energy, or the heat energy storage characteristic of the building envelope structure cannot be fully utilized to absorb the system fluctuation, and the system running cost is high and the energy supply quality is unstable. Disclosure of Invention Aiming at the defects of the prior art, the invention provides a factory production steam and heating ventilation coupling control method based on a neural network, which solves the problem that the characteristics of two different time scales are difficult to effectively process in a unified model frame, so that the prediction result can not accurately reflect the real dynamic response of the system under multiple conditions. The invention aims at achieving the purposes through the following technical scheme that the first aspect of the invention provides a factory production steam and heating ventilation coupling control method based on a neural network, and the method is achieved based on a multi-energy coupling cascade utilization system. The multi-energy coupling cascade utilization system comprises a deep geothermal energy supply loop, an auxiliary steam generating loop and a heating ventilation air conditioning loop which are physically connected. The method mainly comprises the steps of collecting source side parameters, load side parameters and state side parameters of a system, conducting time alignment and standardization processing on collected data to construct a time sequence feature vector reflecti