Method and system for predicting energy consumption of short-process iron and steel enterprises under combined driving of mechanism data

Abstract

The invention relates to a mechanism data combined driven short-process iron and steel enterprise energy consumption prediction method and system, and the method comprises the following steps: solving the continuous rule of the material flow and temperature of the previous and later processes based on the coupling relation of the material flow and the energy flow between the iron and steel production processes, and a metallurgy mechanism; mechanism driving and data driving models are analyzed, and in the data driving model part, an improved hidden semi-Markov model is adopted to fit continuous rules of material flow and temperature of previous and later processes; and constructing an iron and steel enterprise power prediction model combining the mechanism driving model and the data driving model, outputting a power prediction result of each process, carrying out time domain superposition on the power prediction results of each process to obtain an iron and steel enterprise total power characteristic time domain model, and outputting an energy consumption prediction result according to the iron and steel enterprise total power characteristic time domain model.

Description

Technical Field

[0001] The invention relates to a method and system for predicting energy consumption of a short-process steel enterprise driven jointly by mechanism data, and belongs to the technical field of energy consumption prediction of steel enterprises. Background Art

[0002] For predictive research on power characteristics of the steel industry, commonly used methods include mechanism-driven models and data-driven models. Mechanism-driven models are good at analyzing internal mechanisms and extracting hidden knowledge rules. They are highly interpretable, global, and do not rely on data. However, the knowledge mechanism cognition for large-scale complex problems is not clear enough, difficult to model, the cost of acquiring knowledge is high, it is difficult to continuously learn and evolve, and it is impossible to balance calculation accuracy and efficiency. When the mechanism method is used alone, the model complexity is high, and it is inevitable to simplify and omit, resulting in deviations from the actual working conditions on site. The data-driven method does not pay too much attention to the internal mechanism of the research object, but conducts correlation analysis based on a large amount of historical data, and obtains an empirical model after training. However, the data-driven method is overly dependent on the scale and quality of the data, is limited by the data of specific scenarios, and has poor interpretability. The data-driven method has serious "black box" problems, cannot consider the global characteristics of the problem, relies heavily on historical data, and cannot obtain the causal relationship within the data, which restricts its application in actual systems. Summary of the invention

[0003] In order to solve the problems existing in the above-mentioned prior art, the present invention proposes a short-process steel enterprise energy consumption prediction method and system driven by mechanism data, which organically combines the advantages and disadvantages of data-driven and mechanism-driven methods, making the data-driven method not only intelligent but also explainable, which is of great significance for breaking through the current development bottleneck of steel enterprises in power prediction.

[0004] The technical solution of the present invention is as follows:

[0005] On the one hand, the present invention provides a method for predicting energy consumption of a short-process steel enterprise driven by mechanism data, comprising the following steps:

[0006] Based on the coupling relationship between material flow and energy flow in steel production processes and metallurgical mechanism, the continuity law of material flow and temperature in the previous and next processes is solved;

[0007] The mechanism-driven and data-driven models are analyzed. In the data-driven model, an improved hidden semi-Markov model is used to fit the continuous laws of material flow and temperature of the previous and next processes.

[0008] A power prediction model for steel enterprises is constructed by combining the mechanism-driven model and the data-driven model. The mechanism-driven model is used to calculate the attenuation of material flow and energy flow between the previous and next processes. The data-driven model outputs the power prediction results of each process based on the attenuation of material flow and energy flow between the previous and next processes. The power prediction results of each process are superimposed in the time domain to obtain the time domain model of the total power characteristics of the steel enterprise. The energy consumption prediction results are output based on the time domain model of the total power characteristics of the steel enterprise.

[0009] As a preferred embodiment, the step of solving the continuity law of material flow and temperature of the previous and next processes based on the coupling relationship of material flow and energy flow between steel production processes and metallurgical mechanism includes:

[0010] According to the ladle structure and corresponding parameters, the heat flux of the ladle bottom and slag layer, as well as the heat flux of the ladle wall, are obtained by one-dimensional steady-state heat conduction treatment;

[0011] According to the heat flux of the ladle bottom and slag layer, the heat flux of the ladle wall, the heat conduction equation and the constraints of the molten steel transportation time, the linear law of the molten steel temperature and transportation time is obtained.

[0012] As a preferred implementation, in the step of fitting the continuous law of material flow and temperature of the previous and next processes by using the improved hidden semi-Markov model, the material flow and temperature are used as exogenous variables of the hidden semi-Markov model;

[0013] At the same time, the multinomial logistic regression model is used to learn the state transition probability in the improved hidden semi-Markov model;

[0014] The historical power data, material flow and temperature are used as inputs in the learning process of the improved hidden semi-Markov model. The power data is used as the data in the learning process, and the continuous laws of material flow and temperature of the previous and subsequent processes are fitted through iterative training.

[0015] As a preferred implementation, in the step of outputting the energy consumption prediction result according to the time domain model of the total power characteristics of the steel enterprise, additional constraints are set, including:

[0016] The smelting process maintains continuity constraints, the waiting time constraints of the ladle staying outside the equipment, and the power utilization constraints.

[0017] On the other hand, the present invention also provides a mechanism data-jointly driven short-process steel enterprise energy consumption prediction system, comprising:

[0018] The material flow-temperature law solving module solves the continuous law of material flow and temperature of the previous and next processes based on the coupling relationship between material flow and energy flow between steel production processes and metallurgical mechanism;

[0019] The data-driven model building module uses an improved hidden semi-Markov model to build a data-driven model, and the improved hidden semi-Markov model is used to fit the continuous laws of material flow and temperature of the previous and subsequent processes;

[0020] The mechanism-data joint prediction module is used to build a power prediction model for steel enterprises that combines the mechanism-driven model and the data-driven model. The mechanism-driven model is used to calculate the attenuation of material flow and energy flow between the previous and next processes. The data-driven model outputs the power prediction results of each process based on the attenuation of material flow and energy flow between the previous and next processes. The power prediction results of each process are superimposed in the time domain to obtain the time domain model of the total power characteristics of the steel enterprise. The energy consumption prediction results are output based on the time domain model of the total power characteristics of the steel enterprise.

[0021] As a preferred embodiment, the step of solving the continuity law of material flow and temperature of the previous and next processes based on the coupling relationship of material flow and energy flow between steel production processes and metallurgical mechanism includes:

[0022] According to the ladle structure and corresponding parameters, the heat flux of the ladle bottom and slag layer, as well as the heat flux of the ladle wall, are obtained by one-dimensional steady-state heat conduction treatment;

[0023] According to the heat flux of the ladle bottom and slag layer, the heat flux of the ladle wall, the heat conduction equation and the constraints of the molten steel transportation time, the linear law of the molten steel temperature and transportation time is obtained.

[0024] As a preferred implementation, in the step of fitting the continuous law of material flow and temperature of the previous and next processes by using the improved hidden semi-Markov model, the material flow and temperature are used as exogenous variables of the hidden semi-Markov model;

[0025] At the same time, the multinomial logistic regression model is used to learn the state transition probability in the improved hidden semi-Markov model;

[0026] The historical power data, material flow and temperature are used as inputs in the learning process of the improved hidden semi-Markov model. The power data is used as the data in the learning process, and the continuous laws of material flow and temperature of the previous and subsequent processes are fitted through iterative training.

[0027] As a preferred implementation, in the step of outputting the energy consumption prediction result according to the time domain model of the total power characteristics of the steel enterprise, additional constraints are set, including:

[0028] The smelting process maintains continuity constraints, the waiting time constraints of the ladle staying outside the equipment, and the power utilization constraints.

[0029] On the other hand, the present invention also proposes an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for predicting energy consumption of a short-process steel enterprise driven jointly by mechanism data as described in any embodiment of the present invention is implemented.

[0030] On the other hand, the present invention further proposes a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the energy consumption prediction method for a short-process steel enterprise driven jointly by mechanism data as described in any embodiment of the present invention.

[0031] The beneficial effects of the present invention are:

[0032] The present invention discloses a method and system for predicting power of a short-process steel production driven jointly by mechanism and data, which takes into account the coupling relationship between material flow and energy flow in a short-process steel enterprise and is more in line with the actual production process. In the process of fitting the process power curve, two basic parameters of the steel enterprise, namely, mass and temperature, are added to improve the accuracy and dynamics of the model. At the same time, a power consumption prediction model for the steel enterprise combining the mechanism-driven model and the data-driven model is constructed, which can guide the operation optimization of the short-process steel enterprise and can also be used by the steel enterprise to participate in the grid demand response and enhance the market response capability of the enterprise.

[0033] Additional aspects and advantages of the present invention will be set forth in the following description, and some of them will be apparent from the description, or may be understood by practicing the present invention. In addition, the various aspects and advantages of the present invention may be realized and obtained by the method steps and combinations particularly pointed out in the appended claims. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 This is a schematic diagram of a method flow chart of Embodiment 1 of the present invention;

[0035] Figure 2 A schematic diagram of an improved hidden semi-Markov model shown in an embodiment of the present invention;

[0036] Figure 3 This is a schematic diagram of an active power prediction model for a short-process steel enterprise driven jointly by the mechanism and data shown in an embodiment of the present invention. DETAILED DESCRIPTION

[0037] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0038] It should be understood that the step numbers used in this document are only for convenience of description and are not intended to limit the order in which the steps are executed.

[0039] It should be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0040] The terms “include” and “comprising” indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0041] The term "and / or" means and includes any and all possible combinations of one or more of the associated listed items.

[0042] Embodiment 1:

[0043] In view of the technical problems existing in the background technology, this embodiment proposes a short-process power prediction method for steel production driven by mechanism data. First, based on the coupling relationship between material flow and energy flow between processes, the continuity law of material quality and temperature of the previous and next processes is solved in close combination with metallurgical mechanism. Secondly, the relevant models of mechanism drive and data drive are analyzed. In terms of data drive analysis, an improved hidden semi-Markov model is used to fit the power characteristic function based on the coupling characteristics of the physical parameters of the previous and next processes. Finally, the energy consumption prediction results of different processes are superimposed to obtain the time domain model of the total power characteristics of the steel enterprise.

[0044] See also Figure 1 The present embodiment provides a method for predicting energy consumption of a short-process steel enterprise driven by a mechanism data combination, which specifically includes the following steps:

[0045] S100, based on the coupling relationship between material flow and energy flow between steel production processes and metallurgical mechanism, solve the continuity law of material flow and temperature of the previous and next processes;

[0046] S200, analyzing the mechanism-driven and data-driven models. In the data-driven model part, an improved hidden semi-Markov model is used to fit the continuous laws of material flow and temperature of the previous and next processes;

[0047] S300. Construct a power prediction model for steel enterprises that combines a mechanism-driven model and a data-driven model, wherein the mechanism-driven model is used to calculate the attenuation of material flow and energy flow between the previous and next processes, and the data-driven model outputs the power prediction results of each process based on the attenuation of material flow and energy flow between the previous and next processes, and superimposes the power prediction results of each process in the time domain to obtain a time domain model of the total power characteristics of the steel enterprise, and outputs the energy consumption prediction results based on the time domain model of the total power characteristics of the steel enterprise.

[0048] As a preferred implementation of this embodiment, step S100 specifically includes:

[0049] Calculate the temperature loss as follows:

[0050] According to the ladle structure and related parameters, according to the one-dimensional steady-state heat conduction treatment, the heat fluxes of the ladle bottom and slag layer q1 and the ladle wall q2 are as follows:

[0051]

[0052] Where: T1 is the inner surface temperature of the ladle; T f1 is the ambient temperature; i is the thickness of the i-th wall; λ i is the thermal conductivity of the i-th wall; α1 is the radiation heat transfer coefficient; α2 is the convection heat transfer coefficient; r i is the radius of the i-th layer wall; r n+1 is the outer radius of the ladle.

[0053] According to the following heat conduction equation:

[0054] Q loss =c s m·t w ;

[0055] According to the above formula, in formula t w ≤d max Within the constraints, the linear relationship between molten steel temperature and transportation time is:

[0056] T=kt w +b;

[0057] Where: d max The maximum allowable cooling time of the ladle during transportation; t w Indicates the intermediate transportation time. c s is the specific heat of molten steel; Q loss is the heat loss value of molten steel during the intermediate transportation of the ladle; k and b are constants; T is the temperature that changes with t w Varying molten steel temperature.

[0058] The scrap steel produced during the smelting process is transported via a specific ladle. Once the molten steel is poured into the ladle, the ladle will no longer be replaced during the entire production process. Therefore, in the transportation link, the conveying mode remains fixed, and the drop in the temperature of the molten steel is significantly affected by the time factor. Therefore, for the intermediate transportation time:

[0059] t w =t1+t2+t3+t4

[0060] In the formula: t1 represents the conveying time; t2 represents the simple processing time during the conveying process; t3 represents the waiting time, that is, the workstation has products that are processed normally and needs to wait for completion before entering; t4 represents the waiting time due to failures caused by emergencies.

[0061] As a preferred implementation of this embodiment, in step S200, this embodiment adopts the probability statistics method of the improved hidden semi-Markov model to fit the power characteristic function based on the coupling characteristics of the material flow and temperature of the previous and next processes, specifically including the following steps:

[0062] Step S201: Analyze the coupling characteristics of material flow and energy flow within a single process. The product processed in the i-1th process will be transported to the i-th process for further processing with energy. The i-th process reaches the temperature required for processing this process under the drive of external energy (such as electrical energy), and undergoes complex physical and chemical reactions according to process requirements. Under the control of the system, the product produced in this process is transported to the i+1th process for further processing.

[0063] Step S202: Use a machine learning algorithm to establish a fitting curve. The specific method is as follows:

[0064] Compared with the Markov model, the Hidden Semi-Markov Model (HSMM) has the following significant features: it contains hidden unknown parameters and duration, and the state sequence cannot be directly observed. Figure 2 A key feature of the improved hidden semi-Markov model shown is that the state transition probabilities, generation probabilities and duration distributions depend on exogenous variables. This allows factors such as material flow and temperature that affect the physical dynamics of the load in the steel manufacturing process to be incorporated into the model. Material flow (mass) and temperature are treated as exogenous variables, thereby organically combining mechanism-driven models with data-driven models.

[0065] See Figure 2 , the improved hidden semi-Markov model includes two sets of exogenous variables. The first set controls the state and duration transitions between consecutive moments, i.e., the generalized state from moment k-1 At time k The transformation depends on exogenous variables The second group of affected observations is the observed values ​​y t The distribution at time t is based on the current state x t and the exogenous variable w at time t t As a condition. is the temperature, w t It's quality.

[0066] In the time range with T as the period, there are a total of K moments. According to the initial distribution Select the first state and its duration Generalized state According to the probability of generation Produces three observations According to the transition probability Status from Transition to Occupied In the second period, based on Generate six observations This transformation continues until the last observation y T ,y T is the last state of period K This final state May last longer than Long, but set a limited time T.

[0067] In this embodiment, the hidden state transition probability is composed of a The hidden state transition probability matrix A is represented by the hidden state transition probability matrix A. The hidden state transition probability is a set of probability functions. arrive The transition probability is an exogenous variable The function is as follows:

[0068]

[0069] Similarly, the observed value y t With state x t and the exogenous variable w t As a condition.

[0070] b(y t |x t ,w t )=P[y t |x t ,w t ];

[0071] The model does not assume any independence of states or durations.

[0072] In this embodiment, there are five parameters that need to be determined in the model: the number of states N S , duration space D, initial distribution π, hidden state transition probability A and generated observed state probability B. Among them, the number of states can be estimated from the number of significant peaks in the frequency histogram of actual power consumption. Given the number of states, this embodiment applies K-means clustering and then determines the duration space D based on the clustering results. The initial distribution π can be estimated by empirical marginal distribution.

[0073] In order to simplify the estimation of transition probability, this embodiment converts the generalized state arrive The transfer is divided into two parts: state transfer and duration transfer Among them, the generalized state transition is given by:

[0074]

[0075] in:

[0076]

[0077] According to duration analysis, a single distribution is difficult to describe the randomness and correlation of different state durations. Due to the high computational cost, traditional inference algorithms are difficult to estimate a more appropriate conditional distribution for a specific state. In order to solve the computational problem, this embodiment learns the transition probability through multinomial logistic regression (MNLR).

[0078] The MNLR model is used in this embodiment because state and duration are discrete variables. The state of the regression function is used to transform the independent variable Mapped to the state at time k Similarly, use with any duration The associated regression function takes the independent variable Mapped to the duration at time k In particular, the state transition probability and duration transition probability Has the following form:

[0079]

[0080]

[0081] in is the regression coefficient. is a scalar, and and In terms of dimensions, compatible.

[0082] Assume that the generation probability b(y|x,w) is Gaussian distributed. Specifically, assume that the observed value y t is the current state x t and the exogenous variable w at time t t The linear function of , and the residual follows a Gaussian distribution, that is:

[0083] y~N(y x +φ e );

[0084] where γ x is the centroid associated with state x from the K-means clustering algorithm. Based on the states obtained by K-means clustering, the values ​​of parameters α, β, γ, φ, and σ are obtained using maximum likelihood estimation.

[0085] Finally, the improved hidden semi-Markov model constructed above is used for model training. The historical power data, material flow and temperature are used as inputs in the learning process of the improved hidden semi-Markov model. The power data is used as the data in the learning process, and iterative training is performed to fit the continuous laws of material flow and temperature of the previous and subsequent processes.

[0086] In this embodiment, step S300 specifically includes:

[0087] Build as Figure 3 The active power prediction model for short-process steel enterprises driven by the mechanism and data shown in the figure is Figure 3 Each chain q i Representing different processes, these chains achieve coordination, influence and interaction with each other through the synergy of logistics and energy flow, and the attenuation driven by the mechanism becomes the key link connecting the chains; within a single chain, the job arrangement is the link connecting the links, and the data-driven model is used to determine the characteristic parameters of different equipment, and then calculate the entire operation cycle E oc Power variation curve of each device in the system.

[0088] See Figure 3 ,q i represents the i-th process, i∈{1,2,3,4} respectively represents the smelting, refining, continuous casting and steel rolling processes; node is the operating status of the i-th process at time t; E oc Represents an operation cycle, each process i operation undergoes 3 consecutive change states, including the start state Steady state operation and shutdown status r t i-1,iIt represents the material flow and energy flow attenuation between the i-1th process and the ith process at time t, which is calculated by the mechanism-driven method as the sample input value of the data-driven model. In this patent, it is the influence of the material mass and temperature on the active power of the equipment; Δt i represents the waiting time of the ith process, that is, the time the equipment waits for the next task to start after completing one task; Indicates that the device is Active power in state; P t It is the sum of the power of all equipment at time t, that is, the total power of the short-process steel enterprise.

[0089] After the active power prediction model of short-process steel enterprises driven by the joint mechanism and data is built, the power data of different processes at the same time are superimposed in the time domain to obtain the total power of the short-process steel enterprises, as shown in the following formula:

[0090]

[0091] ΔP otherload =(1+Δδ others ) otherrated ;

[0092] Where: N represents the active power of the main equipment i in each process at time t; e Indicates the number of main equipment; ΔP otherload It represents the sum of the power of other loads except the main equipment. Due to its small power fluctuation, it can be regarded as a stable load and does not participate in the subsequent mechanism data joint drive modeling; P otherrated is the sum of the rated powers of other loads; Δδ others is (-δ others ,δ others ), δ others Calculated based on historical statistical values.

[0093] According to the characteristics of the short process of steel enterprises, some additional constraints need to be added in the study of specific problems, including:

[0094] The smelting process is subject to the continuous constraint. For each smelting of molten steel, it can only be processed by a single production equipment at each stage. Once a smelting batch of molten steel begins to be processed, the smelting batch must remain continuous and cannot be suspended before the entire processing flow is completed. The specific constraints are as follows:

[0095]

[0096] Where k is the number of heats, m,n∈{1,2,…,k} and m≠n; Indicates the starting time when the mth heat of molten steel is processed; Indicates the time when the processing of the nth heat of molten steel is completed.

[0097] The waiting time constraint of the ladle staying outside the equipment. In order to ensure that the residence time of the ladle outside the equipment is reasonably controlled and avoid adverse effects on the quality of the ingot, the corresponding waiting time constraint needs to be set, as follows:

[0098] t w ≤d max ;

[0099] Where: d max The maximum allowable cooling time of the ladle during transportation; t w Indicates the intermediate transportation time.

[0100] The power usage constraint means that the power usage at any time cannot exceed the maximum capacity limit of the device. The specific constraints are as follows:

[0101]

[0102] Where: Represents the maximum output power of the i-th process.

[0103] Embodiment 2:

[0104] This embodiment provides a mechanism data-jointly driven short-process steel enterprise energy consumption prediction system, including:

[0105] The material flow-temperature law solving module solves the material flow and temperature continuity law of the previous and next processes based on the coupling relationship between the material flow and energy flow between the steel production processes and the metallurgical mechanism; this module is used to implement the function of step S100 in the first embodiment, which will not be described in detail here;

[0106] A data-driven model building module, which uses an improved hidden semi-Markov model to build a data-driven model, and fits the continuous law of material flow and temperature of the previous and next processes through the improved hidden semi-Markov model; this module is used to implement the function of step S200 in the first embodiment, and will not be repeated here;

[0107] The mechanism-data joint prediction module is used to construct a power prediction model for steel enterprises that combines the mechanism-driven model and the data-driven model, wherein the mechanism-driven model is used to calculate the attenuation of material flow and energy flow between the previous and next processes, and the data-driven model outputs the power prediction results of each process based on the attenuation of material flow and energy flow between the previous and next processes, and the power prediction results of each process are superimposed in the time domain to obtain the time domain model of the total power characteristics of the steel enterprise, and the energy consumption prediction results are output according to the time domain model of the total power characteristics of the steel enterprise; this module is used to implement the function of step S300 in Example 1, which will not be repeated here.

[0108] As a preferred implementation of this embodiment, the steps of solving the continuity law of material flow and temperature of the previous and next processes based on the coupling relationship of material flow and energy flow between steel production processes and metallurgical mechanism include:

[0109] According to the ladle structure and corresponding parameters, the heat flux of the ladle bottom and slag layer, as well as the heat flux of the ladle wall, are obtained by one-dimensional steady-state heat conduction treatment;

[0110] According to the heat flux of the ladle bottom and slag layer, the heat flux of the ladle wall, the heat conduction equation and the constraints of the molten steel transportation time, the linear law of the molten steel temperature and transportation time is obtained.

[0111] As a preferred implementation of this embodiment, in the step of fitting the continuous law of material flow and temperature of the previous and next processes using the improved hidden semi-Markov model, the material flow and temperature are used as exogenous variables of the hidden semi-Markov model;

[0112] At the same time, the multinomial logistic regression model is used to learn the state transition probability in the improved hidden semi-Markov model;

[0113] The historical power data, material flow and temperature are used as inputs in the learning process of the improved hidden semi-Markov model. The power data is used as the data in the learning process, and the continuous laws of material flow and temperature of the previous and subsequent processes are fitted through iterative training.

[0114] As a preferred implementation of this embodiment, in the step of outputting the energy consumption prediction result according to the time domain model of the total power characteristics of the steel enterprise, additional constraints are set, including:

[0115] The smelting process maintains continuity constraints, the waiting time constraints of the ladle staying outside the equipment, and the power utilization constraints.

[0116] Embodiment three:

[0117] This embodiment proposes an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the method for predicting energy consumption of a short-process steel enterprise driven jointly by mechanism data as described in any embodiment of the present invention is implemented.

[0118] Embodiment 4:

[0119] This embodiment proposes a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the method for predicting energy consumption of a short-process steel enterprise driven by the mechanism data as described in any embodiment of the present invention is implemented.

[0120] In the embodiments of the present application, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent the existence of A alone, the existence of A and B at the same time, and the existence of B alone. Among them, A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b and c can be represented by: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, c can be single or multiple.

[0121] Those of ordinary skill in the art will appreciate that the various units and algorithm steps described in the embodiments disclosed herein can be implemented in a combination of electronic hardware, computer software, and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0122] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0123] In several embodiments provided in the present application, if any function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory; hereinafter referred to as: ROM), random access memory (Random Access Memory; hereinafter referred to as: RAM), disk or optical disk, and other media that can store program codes.

[0124] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for predicting energy consumption of a short-process steel enterprise driven by mechanism data, characterized in that: The following steps are involved: Based on the coupling relationship between material flow and energy flow in steel production processes and metallurgical mechanism, the continuity law of material flow and temperature in the previous and next processes is solved; The mechanism-driven and data-driven models are analyzed. In the data-driven model, an improved hidden semi-Markov model is used to fit the continuous laws of material flow and temperature of the previous and next processes. A power prediction model for steel enterprises is constructed by combining the mechanism-driven model and the data-driven model. The mechanism-driven model is used to calculate the attenuation of material flow and energy flow between the previous and next processes. The data-driven model outputs the power prediction results of each process based on the attenuation of material flow and energy flow between the previous and next processes. The power prediction results of each process are superimposed in the time domain to obtain the time domain model of the total power characteristics of the steel enterprise. The energy consumption prediction results are output based on the time domain model of the total power characteristics of the steel enterprise.

2. The method for predicting energy consumption of a short-process steel enterprise driven by mechanism data according to claim 1 is characterized in that: The steps of solving the continuity law of material flow and temperature of the previous and next processes based on the coupling relationship of material flow and energy flow between steel production processes and metallurgical mechanism include: According to the ladle structure and corresponding parameters, the heat flux of the ladle bottom and slag layer, as well as the heat flux of the ladle wall, are obtained by one-dimensional steady-state heat conduction treatment; According to the heat flux of the ladle bottom and slag layer, the heat flux of the ladle wall, the heat conduction equation and the constraints of the molten steel transportation time, the linear law of the molten steel temperature and transportation time is obtained.

3. The method for predicting energy consumption of a short-process steel enterprise driven by a mechanism data combination according to claim 1 is characterized by: In the step of fitting the continuous law of material flow and temperature of the previous and next processes by using the improved hidden semi-Markov model, the material flow and temperature are used as exogenous variables of the hidden semi-Markov model; At the same time, the multinomial logistic regression model is used to learn the state transition probability in the improved hidden semi-Markov model; The historical power data, material flow and temperature are used as inputs in the learning process of the improved hidden semi-Markov model. The power data is used as the data in the learning process, and the continuous laws of material flow and temperature of the previous and subsequent processes are fitted through iterative training.

4. The method for predicting energy consumption of a short-process steel enterprise driven by mechanism data according to claim 1 is characterized in that: In the step of outputting the energy consumption prediction results based on the time domain model of the total power characteristics of the steel enterprise, additional constraints are set, including: The smelting process maintains continuity constraints, the waiting time constraints of the ladle staying outside the equipment, and the power utilization constraints.

5. A mechanism data-jointly driven short-process steel enterprise energy consumption prediction system, characterized in that: include: The material flow-temperature law solving module solves the continuous law of material flow and temperature of the previous and next processes based on the coupling relationship between material flow and energy flow between steel production processes and metallurgical mechanism; The data-driven model building module uses an improved hidden semi-Markov model to build a data-driven model, and the improved hidden semi-Markov model is used to fit the continuous laws of material flow and temperature of the previous and subsequent processes; The mechanism-data joint prediction module is used to build a power prediction model for steel enterprises that combines the mechanism-driven model and the data-driven model. The mechanism-driven model is used to calculate the attenuation of material flow and energy flow between the previous and next processes. The data-driven model outputs the power prediction results of each process based on the attenuation of material flow and energy flow between the previous and next processes. The power prediction results of each process are superimposed in the time domain to obtain the time domain model of the total power characteristics of the steel enterprise. The energy consumption prediction results are output based on the time domain model of the total power characteristics of the steel enterprise.

6. The energy consumption prediction system for short-process steel enterprises driven by mechanism data combination according to claim 5 is characterized in that: The steps of solving the continuity law of material flow and temperature of the previous and next processes based on the coupling relationship of material flow and energy flow between steel production processes and metallurgical mechanism include: According to the ladle structure and corresponding parameters, the heat flux of the ladle bottom and slag layer, as well as the heat flux of the ladle wall, are obtained by one-dimensional steady-state heat conduction treatment; According to the heat flux of the ladle bottom and slag layer, the heat flux of the ladle wall, the heat conduction equation and the constraints of the molten steel transportation time, the linear law of the molten steel temperature and transportation time is obtained.

7. The energy consumption prediction system for short-process steel enterprises driven by mechanism data combination according to claim 5 is characterized by: In the step of fitting the continuous law of material flow and temperature of the previous and next processes by using the improved hidden semi-Markov model, the material flow and temperature are used as exogenous variables of the hidden semi-Markov model; At the same time, the multinomial logistic regression model is used to learn the state transition probability in the improved hidden semi-Markov model; The historical power data, material flow and temperature are used as inputs in the learning process of the improved hidden semi-Markov model. The power data is used as the data in the learning process, and the continuous laws of material flow and temperature of the previous and subsequent processes are fitted through iterative training.

8. The energy consumption prediction system for short-process steel enterprises driven by mechanism data combination according to claim 5 is characterized in that: In the step of outputting the energy consumption prediction results based on the time domain model of the total power characteristics of the steel enterprise, additional constraints are set, including: The smelting process maintains continuity constraints, the waiting time constraints of the ladle staying outside the equipment, and the power utilization constraints.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the method for predicting energy consumption of a short-process steel enterprise driven by the mechanism data as described in any one of claims 1 to 4 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method for predicting energy consumption of a short-process steel enterprise driven by the mechanism data as described in any one of claims 1 to 4 is implemented.

Images