A multi-energy complementary clean energy supply system based on load prediction for an oil and gas field station

Abstract

The application relates to an oil and gas field station multi-energy complementary clean energy supply system based on load prediction, which comprises: a station energy terminal, which is used for refrigeration / heating of an oil and gas field station through electric energy; a multi-energy complementary energy supply module, which is used for directly supplying the station energy terminal with standard green electric energy converted from solar energy and wind energy, and storing excess standard green electric energy into a lithium battery when the real-time conversion amount of the standard green electric energy is greater than the real-time power load of the station energy terminal; a multi-source heat pump coupling process module, which is used for refrigeration / heating of the oil and gas field station through low-grade heat energy; and an intelligent control module, which is used for predicting a load prediction value, controlling the lithium battery to supply power to the multi-source heat pump coupling process module, and cooperatively controlling the start and stop of the multi-source heat pump coupling process module. The application can accurately predict the load of the oil and gas field station, construct a clean energy supply system, realize supply-demand matching, reduce carbon and improve efficiency, and guarantee stable energy supply.

Description

Technical Field

[0001] This application relates to the field of combined heating and cooling systems, and more particularly to a multi-energy complementary clean energy supply system for oil and gas field stations based on load forecasting. Background Technology

[0002] Oil and gas resources are a crucial component of my country's energy strategy. Oil and gas field stations, as the core sites for oil and gas resource extraction, gathering, transportation, and processing, have a continuous and stable rigid demand for heating and cooling during their production operations, making them typical high-energy-consuming industrial sectors. With the deepening of global carbon neutrality goals and increasingly stringent environmental regulations, the pressure to transform traditional high-carbon energy supply models continues to increase. Optimizing the energy structure and achieving clean and efficient energy supply have become core demands for the sustainable development of the oil and gas field industry. At the same time, oil and gas field areas generally possess abundant clean energy resources such as solar, wind, and geothermal energy, and the production process generates a large amount of underutilized energy, providing a good resource foundation for building clean energy supply systems.

[0003] In existing technologies, oil and gas field stations primarily rely on direct power and heat supply from fossil fuels such as coal and oil. While some stations have attempted to connect solar photovoltaic and wind power generation equipment, this merely involves directly connecting the generated green electricity to the station's heating / cooling system. Because green electricity production is significantly affected by natural conditions such as sunlight and wind, exhibiting inherent intermittency and volatility, direct connection to the energy supply system easily leads to fluctuating heating / cooling power outputs, failing to meet the station's rigid requirements for energy supply stability. Furthermore, the lack of load forecasting methods makes it difficult to anticipate demand changes in energy dispatch, further exacerbating supply-demand mismatch. When green electricity is surplus, it cannot be effectively stored and is wasted; when green electricity is insufficient, emergency replenishment from fossil fuels is necessary, failing to fully realize the environmental benefits of clean energy supply.

[0004] Therefore, it is necessary to improve one or more of the problems existing in the above-mentioned related technical solutions.

[0005] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] The purpose of this disclosure is to provide a multi-energy complementary clean energy supply system for oil and gas field stations based on load forecasting, thereby overcoming, to at least some extent, one or more problems caused by the limitations and defects of related technologies.

[0007] This application provides a multi-energy complementary clean energy supply system for oil and gas field stations based on load forecasting, including: Station energy terminals are used to cool / heat oil and gas field stations using electrical energy; The multi-energy complementary power supply module is used to convert solar and wind energy into standard green electricity and then directly supply it to the station's energy consumption terminal. When the real-time conversion amount of standard green electricity is greater than the real-time power load of the station's energy consumption terminal, the excess standard green electricity is stored in the lithium battery. The multi-source heat pump coupling process module includes a ground source heat pump unit, an air source heat pump unit, a heat collection circulation unit, and a load-side water tank circulation unit. It collects low-grade heat energy from the oil and gas field station through the ground source heat pump unit, air source heat pump unit, and heat collection circulation unit. When powered by a lithium battery, it converts the low-grade heat energy into high-grade cold / heat energy through a reverse Carnot cycle. The cold / heat medium is then transported to the oil and gas field station for cooling / heating via the load-side water tank circulation unit. The low-grade heat energy includes ground source heat energy, air source heat energy, and water source heat energy. The intelligent control module includes a data acquisition unit, a data filtering unit, a model building unit, a load forecasting unit, and an energy control unit. The data acquisition unit is used to collect historical energy consumption data of oil and gas field stations and corresponding historical regional meteorological data, and to perform data cleaning, standardization, normalization, and outlier correction to generate a high-quality basic dataset. The data filtering unit is used to perform load participation factor filtering, correlation analysis and feature classification filtering on the high-quality basic dataset to construct a load prediction feature set; The model building unit is used to build a multi-task learning prediction model based on a long short-term memory network, and to train the model using the load prediction feature set to obtain an MTL-LSTM model. The load forecasting unit is used to collect energy consumption data and regional meteorological data of oil and gas field stations in real time. After data preprocessing and feature screening, the data is input into the MTL-LSTM model to obtain preliminary forecast values. After weight optimization and linear weighted fusion, the predicted values ​​of electricity, cooling and heating loads of the station's energy consumption terminals are obtained. The power control unit is used to control the lithium battery to supply power to the multi-source heat pump coupling process module and the station energy terminal based on the predicted values ​​of the electric, cooling, and heating loads when the predicted values ​​of the electric, cooling, and heating loads are greater than the standard green energy real-time conversion amount, so as to realize the supplementary cooling / heating of the oil and gas field station by the multi-source heat pump coupling process module.

[0008] The technical solution provided in this application may include the following beneficial effects: This application presents a multi-energy complementary clean energy supply system for oil and gas field stations based on load forecasting. Through multi-stage data preprocessing, multi-dimensional feature screening, and MTL-LSTM multi-task learning model training, it achieves accurate forecasting of the electricity, cooling, and heating loads of oil and gas field stations. Furthermore, relying on multi-energy complementary energy supply modules, it efficiently converts solar and wind energy into green electricity, prioritizing direct supply to station energy terminals and storing surplus electricity in lithium batteries. Combined with a multi-source heat pump coupling process module, it recovers and utilizes low-grade heat energy from ground, air, and water sources, converting it into high-grade cooling / heating energy driven by the green electricity stored in lithium batteries to meet the station's combined cooling and heating needs. Simultaneously, through an intelligent control module, it dynamically controls the direction of lithium battery power delivery and the operating status of multi-source heat pumps based on load forecasting results, achieving precise matching of energy supply and demand, ensuring continuous and stable energy supply, significantly reducing fossil fuel consumption and carbon emissions at oil and gas field stations, and improving energy utilization efficiency and supply reliability.

[0009] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0010] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure. It is obvious that the drawings described below are merely some embodiments of this disclosure, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0011] Figure 1 This diagram illustrates the structure of a multi-energy complementary clean energy supply system for oil and gas field stations based on load forecasting, as shown in an exemplary embodiment of this disclosure. Figure 2 This diagram illustrates the data acquisition unit structure of the intelligent control module of the multi-energy complementary clean energy supply system for oil and gas field stations based on load forecasting in an exemplary embodiment of this disclosure. Figure 3 This diagram illustrates the data filtering unit structure of the intelligent control module of the oil and gas field station multi-energy complementary clean energy supply system based on load forecasting in an exemplary embodiment of this disclosure. Figure 4 This diagram illustrates the model building unit structure of the intelligent control module for a multi-energy complementary clean energy supply system for oil and gas field stations based on load forecasting in an exemplary embodiment of this disclosure. Figure 5 This diagram illustrates the load forecasting unit structure of the intelligent control module of the oil and gas field station multi-energy complementary clean energy supply system based on load forecasting in an exemplary embodiment of this disclosure. Figure 6 This diagram illustrates the structure of a multi-energy complementary clean energy supply module in an exemplary embodiment of this disclosure. Figure 7 This diagram illustrates the structure of a multi-source heat pump coupling process module in an oil and gas field station multi-energy complementary clean energy supply system based on load forecasting, as shown in an exemplary embodiment of this disclosure. Figure 8 This diagram illustrates the structure of the intelligent control module of the oil and gas field station multi-energy complementary clean energy supply system based on load forecasting in an exemplary embodiment of this disclosure. Detailed Implementation

[0012] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that this disclosure will be more comprehensive and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0013] This example implementation first provides a multi-energy complementary clean energy supply system for oil and gas field stations based on load forecasting. (Reference) Figure 1 As shown, the system may include a station energy terminal, a multi-energy complementary energy supply module, a multi-source heat pump coupling process module, and an intelligent control module.

[0014] Among them, the station energy terminal is used to cool / heat the oil and gas field station using electricity; The multi-energy complementary power supply module is used to convert solar and wind energy into standard green electricity and then directly supply it to the station's energy consumption terminal. When the real-time conversion amount of standard green electricity is greater than the real-time power load of the station's energy consumption terminal, the excess standard green electricity is stored in the lithium battery. The multi-source heat pump coupling process module includes a ground source heat pump unit, an air source heat pump unit, a heat collection circulation unit, and a load-side water tank circulation unit. It collects low-grade heat energy from the oil and gas field station through the ground source heat pump unit, air source heat pump unit, and heat collection circulation unit. When powered by a lithium battery, it converts the low-grade heat energy into high-grade cold / heat energy through a reverse Carnot cycle. The cold / heat medium is then transported to the oil and gas field station for cooling / heating via the load-side water tank circulation unit. The low-grade heat energy includes ground source heat energy, air source heat energy, and water source heat energy. The intelligent control module includes a data acquisition unit, a data filtering unit, a model building unit, a load forecasting unit, and an energy control unit. The data acquisition unit is used to collect historical energy consumption data and corresponding historical regional meteorological data from oil and gas field stations, and to perform data cleaning, standardization, normalization, and outlier correction to generate a high-quality basic dataset. The energy consumption data includes electrical load operation data, cooling load operation data, and heating load operation data, and the regional meteorological data includes ambient temperature, solar radiation, and air humidity. The data filtering unit is used to perform load participation factor filtering, correlation analysis and feature classification filtering on the high-quality basic dataset to construct a load prediction feature set; The model building unit is used to build a multi-task learning prediction model based on a long short-term memory network, and to train the model using the load prediction feature set to obtain an MTL-LSTM model. The load forecasting unit is used to collect energy consumption data and regional meteorological data of oil and gas field stations in real time. After data preprocessing and feature screening, the data is input into the MTL-LSTM model to obtain preliminary forecast values. After weight optimization and linear weighted fusion, the predicted values ​​of electricity, cooling and heating loads of the station's energy consumption terminals are obtained. The power control unit is used to control the lithium battery to supply power to the multi-source heat pump coupling process module and the station energy terminal based on the predicted values ​​of the electric, cooling, and heating loads when the predicted values ​​of the electric, cooling, and heating loads are greater than the standard green energy real-time conversion amount, so as to realize the supplementary cooling / heating of the oil and gas field station by the multi-source heat pump coupling process module.

[0015] It should be noted that the predicted values ​​of electrical load, cooling load, and heating load are respectively the electrical load for power supply, cooling load, and heating load of the station's energy-consuming terminals; the standard green energy is the electricity generated by solar power generation equipment and wind power generation equipment through multi-energy complementary power supply modules. The station's energy terminals can be cooling / heating equipment for operating areas, dormitory areas, and living areas within the oil and gas field station, as well as other cooling / heating equipment for areas requiring cooling / heating.

[0016] Data transmission between the various modules of the system can adopt the Modbus-RTU communication protocol to ensure the stability and compatibility of data interaction. The core logic of the system operation is load forecasting-driven energy dispatch, that is, based on the predicted electricity, cooling and heating load demand, dynamically adjusting the output of wind and solar power generation, the charging and discharging status of lithium batteries and the operation mode of heat pumps to achieve precise matching of energy supply and demand.

[0017] The low-grade thermal energy of the oil and gas field stations includes: air thermal energy, which is low-grade thermal energy that is widely present in the atmospheric environment and is extracted through finned heat exchangers; shallow geothermal energy, which originates from soil or rock several to hundreds of meters underground and has a relatively constant temperature and can be obtained through geothermal tube heat exchangers; groundwater and surface water source thermal energy, including the heat contained in rivers, lakes and stable water sources around the oil field; and production-related waste heat unique to oil and gas fields, such as the heat carried by crude oil produced fluid, oily wastewater or gas well fluid during separation and treatment.

[0018] The reverse Carnot cycle, based on the thermodynamic reverse cycle theory, consumes a small amount of high-grade electrical energy—specifically, the energy stored in lithium batteries—to drive the refrigerant through a state change within a closed system consisting of four core components: a compressor, a condenser, an expansion valve, and an evaporator. This achieves the reverse transfer of heat from a low-temperature heat source to a high-temperature heat source. Its high efficiency is reflected in its energy efficiency ratio, typically reaching 3 to 5, meaning that consuming 1 unit of electrical energy can transfer 3 to 5 units of heat energy. The specific working mechanism is as follows: In heating mode, the low-temperature, low-pressure liquid refrigerant first absorbs heat from a low-grade heat source in the evaporator, vaporizing into low-temperature, low-pressure steam. This steam is then drawn into and compressed by the compressor, becoming high-temperature, high-pressure superheated steam. This high-temperature steam enters the condenser, where it exchanges heat with the circulating water in the oil and gas field heating system. The steam condenses and releases heat, heating the circulating water to the required temperature for external heating. Finally, the high-pressure liquid refrigerant is throttled and depressurized by the expansion valve, returning to a low-temperature, low-pressure liquid state and re-entering the evaporator to absorb heat, thus completing the cycle. The cooling mode is achieved by changing the refrigerant flow direction through a four-way reversing valve. In this mode, the indoor heat exchanger acts as an evaporator to absorb heat and cool, while the outdoor heat exchanger acts as a condenser to release heat to the environment.

[0019] Regarding the specific circulation and application process of hot / cold water in oil and gas fields, this process constitutes a closed-loop circulation system with heat pumps at its core and water as the heat / cold medium. Specifically, after heat exchange is completed in the condenser (for heating) or evaporator (for cooling) of the multi-source heat pump unit, the resulting hot (or cold) water, meeting the process temperature requirements, is pressurized by a circulating water pump and transported to different energy-consuming terminals in the oil and gas field via insulated pipelines laid in trenches or overhead. For example, high-temperature hot water is mainly input into the wellhead heating coils, water mixing pipelines, or heating coils of crude oil storage tanks in the oil and gas gathering and transportation system. Through indirect heat exchange, heat is transferred to the produced fluid in the wellbore or the crude oil in the storage tank to reduce its viscosity, ensure fluidity, prevent condensation accidents, and meet the process heat requirements for continuous production. Low-temperature cold water is input into the air conditioning terminal systems of oil and gas field auxiliary buildings, such as office buildings and central control rooms, such as fan coil units; or into specific process stages, such as cooling before compression of certain gases, for summer cooling. After hot (or cold) water releases heat (or cooling) at the user's end, its temperature decreases (or increases), becoming "return water." This return water flows back to the heat pump unit through an independent return water network, relying on the suction of a circulating pump or gravity. After re-entering the heat pump unit, the return water will undergo heat exchange again in the evaporator (during heating) or condenser (during cooling), achieving cascaded heat recovery and recycling.

[0020] It is understandable that the reason for using multi-source heat pump coupling process modules to supplement the energy supply to the station's energy terminals when electricity demand is too high is as follows: Based on the principle of energy cascade utilization according to the second law of thermodynamics, wind and solar power generation is high-grade electrical energy, which can directly drive electric air conditioning or electric heating equipment to achieve direct conversion with "temperature matching and energy level matching". If high-grade electrical energy is input into a multi-source heat pump system to drive the compressor, although the heat pump has a higher power utilization rate, it is essentially using electrical energy to transport low-grade environmental heat energy, adding multiple intermediate links in the energy conversion chain, such as compressor losses, water pump distribution losses, and pipeline heat losses.

[0021] From the perspective of overall system economy, the priority direct supply solution only requires the configuration of wind and solar power generation equipment and a small number of multi-source heat pump units as supplements. The cost of a single multi-source heat pump unit is about 1.15 million yuan and its service life is only 10 years, with high replacement and maintenance costs. On the other hand, the solution of supplying all green electricity to the heat pump requires the configuration of more heat pump units, resulting in a significant increase in initial investment. Moreover, the contradiction between the volatility of green electricity and the continuous operation requirements of heat pumps will give rise to additional energy storage investment, and the economic efficiency will be significantly deteriorated.

[0022] From the perspective of grid stability and the safety of green energy consumption, wind and solar power generation exhibits significant intermittent and fluctuating characteristics, with their output curves displaying random fluctuations. In the priority direct supply scheme, decentralized electric air conditioning or electric heating equipment, acting as flexible loads, can adapt their power consumption to real-time changes in wind and solar output. Voltage and frequency fluctuations can be adaptively adjusted at the equipment level, without causing a shock to the distribution network. This is a safe and stable way to maximize the consumption of green energy. Conversely, if fluctuating green energy is directly injected into a centralized multi-source heat pump system, the heat pump unit, as a rigid load, has strict requirements for power quality. Voltage fluctuations can easily trigger compressor protection shutdowns and inverter failures, and in severe cases, may lead to mechanical damage to the compressor, significantly reducing the reliability of the system's power supply and the service life of the equipment.

[0023] From the perspective of ensuring energy supply reliability and system redundancy, oil and gas field production and living areas have extremely high requirements for the continuity of heating and cooling. The priority direct supply solution relies on tens of millions of independently operating household air conditioners or heating devices to form a distributed energy supply architecture. The failure of a single device only affects a local area, and the overall system redundancy is high, which conforms to the design concept of fault isolation and minimizing impact. On the other hand, multi-source heat pumps, as centralized heat and cold sources, rely entirely on the integrity of a single main unit, circulating water pump, and pipeline system for operational reliability. Once the core equipment fails or the pipeline leaks, it will lead to the interruption of energy supply to the entire living area, creating a single point of failure risk, which is difficult to meet the energy supply security requirements of oil and gas field scenarios.

[0024] From the perspective of dynamic response characteristics and load matching capabilities, the heating and cooling loads in residential areas exhibit significant instantaneous changes, showing a step-like pattern of sudden increases during the evening peak and sharp decreases during the day. Electric air conditioning or electric heating equipment uses direct expansion or resistance heat exchange, with start-up and shutdown response times within the second range, which can effectively match the instantaneous fluctuations in wind and solar power output and the rapid changes in load demand. However, multi-source heat pump systems are high-inertia thermodynamic systems, and their operation requires multiple stages such as compressor start-up, refrigerant circulation establishment, and water temperature gradient rise. Their dynamic response time is relatively long, making it difficult to keep up with the high-frequency fluctuations of wind and solar power. When green electricity output drops sharply, the heat pump system cannot quickly reduce its load, easily leading to energy waste or oversupply; when output surges, the heat pump cannot promptly increase the load, resulting in power curtailment, and the system's regulation capability is severely lagging.

[0025] In summary, the dispatching strategy of prioritizing direct supply of wind and solar green electricity to cooling and heating terminals, with multi-source heat pumps serving as supplementary safeguards, organically combines the direct utilization of high-grade energy with the leverage effect of low-grade environmental energy. This scheme adheres to the principle of cascaded utilization in the energy conversion chain, significantly reduces the total life-cycle cost in terms of investment and operation and maintenance, effectively mitigates fluctuations in grid security, and constructs a highly redundant distributed architecture for energy supply security. In the design of multi-energy complementary systems, wind and solar power can be prioritized for connection to distributed air conditioning or heating loads, utilizing their rapid response characteristics to achieve local consumption of green electricity. Simultaneously, a multi-source heat pump system of appropriate capacity can be configured to handle the basic load or provide supplementary safeguards. Through time-series complementarity, a dynamic balance between energy supply and demand is achieved, providing technical support for the green and low-carbon transformation of oil and gas fields.

[0026] In one embodiment, such as Figure 6 As shown, the multi-energy complementary power supply module includes: a photovoltaic power generation unit, a wind power generation unit, a voltage regulator and boost circuit, a lithium battery energy storage unit, and an energy detection unit; The power output terminals of the photovoltaic power generation unit and the wind power generation unit are connected to the input terminal of the voltage regulator and boost circuit. The output terminals of the voltage regulator and boost circuit are respectively connected to the station's energy consumption terminal and the charging terminal of the lithium battery energy storage unit; The power detection unit is connected to the photovoltaic power generation unit, the wind power generation unit and the station energy terminal respectively, and is used to detect the wind power generation, photovoltaic power generation and the energy load of the station energy terminal in real time. The surface of the lithium battery energy storage unit is covered with graphene-enhanced phase change thermal storage material, and the discharge end is connected to the power input end of the station energy terminal and the multi-source heat pump coupling process module, respectively.

[0027] It should be noted that the output voltage of the voltage regulator and boost circuit can be 380V three-phase AC, adapting to the input voltage requirements. The DC output from the photovoltaic power generation unit and the wind power generation unit can be converted from DC to AC through an inverter circuit. The capacity of the lithium battery energy storage unit can be configured according to the power consumption of the station's maximum load for 2 hours. The SOC full charge threshold can be set to 95%, and the lower limit threshold can be set to 20% to avoid overcharging and over-discharging, which would lead to battery life degradation. The phase change temperature of the graphene-reinforced phase change thermal storage material can be 30℃-35℃, and the latent heat of phase change is ≥180kJ / kg. The latent heat itself enables adaptive temperature regulation of the lithium battery, ensuring that the battery operating temperature is maintained within the range of 20℃-40℃. Simultaneously, a mains backup interface can be configured to maintain the station's cooling and heating supply in the event of unexpected situations such as green power supply interruptions.

[0028] In one embodiment, such as Figure 7 As shown, the solar collector circulation unit includes a PVT collector, a collector circulation pump, and a hot water storage tank. The PVT collector is connected to the hot water storage tank through the collector circulation pump to form a solar collector circuit, which is used to collect low-grade solar thermal energy and store it in the hot water storage tank. The hot water storage tank is connected to the ground source heat pump unit and the air source heat pump unit respectively, and is used to deliver the stored low-grade thermal energy to the ground source heat pump unit and the air source heat pump unit. The ground source heat pump unit includes a first compressor, a first evaporator, a first condenser, and a first throttling valve, which are connected in sequence to form a first closed refrigerant circulation loop; the first evaporator is connected to the hot water storage tank and is used to collect low-grade heat energy from the ground source and low-grade heat energy in the hot water storage tank; the first condenser is connected to the load-side water tank circulation unit and is used to output high-grade heat energy / cold energy after heat pump circulation conversion. The air source heat pump unit includes a second compressor, a second evaporator, a second condenser, and a second throttling valve, which are connected in sequence to form a second closed refrigerant circulation loop; the second evaporator is connected to the hot water storage tank and is used to collect low-grade heat energy from the air source and low-grade heat energy from the hot water storage tank; the second condenser is connected to the load-side water tank circulation unit and is used to output high-grade heat energy / cold energy after heat pump circulation conversion. The load-side water tank circulation unit includes a load-side water tank, a load-side water tank circulation pump, and a terminal circulation pump. The load-side water tank is connected to the condenser outlets of the ground source heat pump unit and the air source heat pump unit via the load-side water tank circulation pump, respectively, for receiving and storing high-grade heat / cold energy output by the ground source heat pump unit and the air source heat pump unit. The outlet of the load-side water tank is connected to the energy-consuming terminal of the oil and gas field station via the terminal circulation pump, for transporting cold / heat medium to the energy-consuming terminal to achieve energy supply.

[0029] It should be noted that the control execution layer uses an STM32 series microcontroller as the main controller. Based on the energy dispatch strategy, it controls the on / off switching of the charging and discharging circuit of the lithium battery energy storage unit to regulate the direction of power transmission, and simultaneously sends start / stop, load adjustment, and operating mode switching commands to the multi-source heat pump coupling process module. The human-machine interaction layer includes an OLED display panel, an LD3320 voice recognition module, and an ESP8266 WiFi module, used to realize local operation data display, voice control, and remote monitoring and command issuance via a mobile terminal APP.

[0030] In one embodiment, the energy-consuming terminal is a finned heat exchanger used for cooling / heating via the cold / hot medium supplied to its interior.

[0031] It should be noted that this finned heat exchanger adopts a corrosion-resistant alloy material and a variable-pitch fin structure design, which can not only adapt to the harsh working conditions of high salt spray and high dust in oil and gas field stations, but also enhance the heat exchange efficiency between the cold / hot medium and the ambient air by increasing the heat exchange area; at the same time, it can be connected to the system load prediction module to dynamically adjust the medium flow rate and temperature according to the real-time cold and hot load demand, avoid redundant energy consumption, and achieve the synergy of accurate energy supply and optimal energy efficiency.

[0032] In one embodiment, the data acquisition unit includes: a historical data acquisition element, a historical data cleaning element, and a historical data processing element.

[0033] It should be noted that, optionally, the data collection time span must cover at least three full years to ensure that energy consumption characteristics under different seasons and weather conditions are included, and the sampling frequency is no less than once per minute to capture the instantaneous fluctuation patterns of the load.

[0034] The historical data acquisition element is used to collect historical data on the electricity, cooling, and heating loads of oil and gas field stations in a time series manner, and simultaneously collect the corresponding historical station area ambient temperature, solar radiation, and air humidity to form the original historical dataset.

[0035] It should be noted that the electrical load data comes from the station's power monitoring module, while the cold / heat load data can be calculated using the supply and return water temperature difference and flow sensor on the load side water tank. That is, cold / heat load = medium density × specific heat capacity × flow rate × supply and return water temperature difference. Meteorological data is preferentially collected from the station's self-built meteorological station. If there is no self-built meteorological station, data from a third-party meteorological platform within 10km of the station can be used. All data must be accurately synchronized through timestamps to ensure that the energy consumption data and meteorological data correspond one-to-one at the same time.

[0036] The historical data cleaning component is used to initially screen out abnormal data that exceeds the reasonable value range using the statistical threshold method. It then performs secondary verification using the time series trend method and the correlation method to distinguish between single-point anomalies and continuous short-term anomalies. Finally, it corrects the abnormal values ​​by using the adjacent time effective value replacement method and the same period and working condition data replacement method, respectively, thus completing the data cleaning.

[0037] It should be noted that the statistical threshold method is based on the 3σ principle, that is, data exceeding the range of [μ-3σ,μ+3σ] is judged as abnormal, where μ is the data mean and σ is the standard deviation; the judgment standard of the time series trend method is that the data change rate between adjacent time points does not exceed 50%, and if it exceeds, it is judged as abnormal; the same period and same working condition data refers to the historical load data of the same season, the same production shift, and similar meteorological conditions. Similar meteorological conditions can be environmental temperature deviation ≤2℃ and solar radiation deviation ≤100W / m²; abnormal data in key periods can be replaced by model prediction, such as the winter heating peak and the summer cooling peak. The prediction model uses a linear regression model, and the input features are the same period meteorological data and production working condition data.

[0038] Historical data processing components are used to standardize and normalize the cleaned dataset to obtain a high-quality basic dataset.

[0039] It should be noted that standardization requires unifying all data into a single unit, i.e., electricity, cooling, and heating loads are all measured in kW, ambient temperature in °C, and solar radiation in W / m². 2 The air humidity is %; normalization is performed using the Min-Max method, the mapping interval is [0,1], and the calculation formula is x'=(xx min ) / (x max -x min ), where x is the original data, x min x max These are the minimum and maximum values ​​for this type of data, ensuring that data of different dimensions can be used for model training.

[0040] In one embodiment, the data filtering unit includes: a factor filtering element, a correlation analysis element, and a feature filtering element.

[0041] It should be noted that by using a progressive process of "load participation factor screening - correlation analysis - feature hierarchical screening", a load prediction feature set adapted to model training is constructed to ensure that the features input to the model not only conform to the energy consumption patterns of oil and gas field stations, but also reduce the computational complexity of the model.

[0042] The factor screening element is used to screen the load participation factors of the high-quality basic dataset. The screening rule for the load participation factors is to determine whether the obtained load participation factors can effectively quantify the correlation between the sub-items of the load and the total load, and whether the contribution of the core load reaches the conventional energy consumption level of the oil and gas field station.

[0043] It should be noted that "the contribution of core loads reaches the level of conventional energy consumption at oil and gas field stations" specifically means that the load participation factor (LPF) value of core loads is not less than 0.4, and the LPF value of auxiliary loads is not less than 0.1. If the LPF value of a certain type of load is lower than the corresponding threshold, it indicates that its impact on the total load is minimal, and only the load characteristics that meet the LPF value standard and the corresponding associated meteorological characteristics are retained. Core loads include wellhead heating, crude oil transportation heat tracing, and crude oil storage tank insulation, while auxiliary loads include office building air conditioning, lighting, and power supply for non-core equipment.

[0044] The formula for calculating the load participation factor is as follows: in, For the first Participating factors of class load, Number of load types For the first , For the first The correlation between the load class and the total load.

[0045] It should be noted that the sum of LPF is 1, and its function is to quantify the proportion and weight of a single type of load in the total load. The higher the weight, the greater the impact of the load on the total load, and it should be included as a key feature in the subsequent correlation analysis.

[0046] The correlation analysis element is used to calculate the correlation between electrical, cooling, and heating load characteristics, as well as between each load characteristic and meteorological characteristics such as ambient temperature, solar radiation, and air humidity, after the load participation factors are screened, using the grey relational analysis method, and obtain the correlation values ​​between each characteristic.

[0047] It should be noted that this grey relational analysis first performs dimensionless preprocessing on the original load and meteorological data to eliminate the interference of dimensional differences on the calculation results. Then, by calculating the correlation coefficient and correlation degree sequence, the degree of correlation between each feature is ranked, and the core factors that have the most significant impact on electricity, cooling and heating loads are screened out. This provides a quantitative basis for the feature engineering and model structure optimization of the subsequent load forecasting model, thereby improving the accuracy and robustness of load forecasting.

[0048] The formula for calculating the degree of correlation is as follows: in, To quantify the correlation of individual data points, As a reference sequence, For comparing sequences, The resolution coefficient, This represents the global maximum value of the differences between all comparison sequences and the reference sequence at each time step. This represents the global minimum of the differences between all comparison sequences and the reference sequence at each time step. The formula for calculating the correlation degree is: in, For relevance, is the length of the time series, and k is the time index of the time series.

[0049] It should be noted that the resolution coefficient ρ of the grey relational analysis can range from 0.1 to 0.5. A value of 0.5 can balance the differences in the degree of correlation between different features. When calculating, the total load sequence is used as the reference sequence, and the individual load sequences of electricity / cooling / heating and the meteorological factor sequences of ambient temperature / solar radiation / air humidity are used as comparison sequences. The degree of correlation between each pair is calculated. The closer the degree of correlation is to 1, the stronger the nonlinear correlation between the two.

[0050] In step S230, the feature filtering element is used to perform feature classification filtering based on the correlation degree value to obtain a load prediction feature set; wherein, features with correlation degree > 0.6 are defined as strongly correlated core features, features with correlation degree 0.3 ≤ correlation degree ≤ 0.6 are defined as moderately correlated core features, and features with correlation degree < 0.3 are defined as weakly correlated redundant features, thus obtaining a load prediction feature set.

[0051] It should be noted that all the graded features together constitute the load prediction feature set, providing comprehensive feature input support for subsequent model training.

[0052] In one embodiment, the model building unit includes: architecture building elements, model training elements, and model optimization elements.

[0053] The architecture building blocks are used to construct a multi-task learning and prediction model architecture based on a long short-term memory network; the model includes a shared LSTM layer, a task-specific LSTM layer, an attention layer, a dropout layer, and an output layer.

[0054] It should be noted that the multi-task learning architecture can simultaneously learn the common patterns and unique characteristics of the three types of loads: electricity, cooling, and heat, avoiding information fragmentation caused by separate modeling and improving prediction efficiency and consistency; model training can be performed on servers with GPU acceleration to ensure training efficiency.

[0055] The model training element is used to input the load prediction feature set into the MTL-LSTM model, extract the temporal common patterns of features through the shared LSTM layer, and then learn the individual patterns of single loads through the task-specific LSTM layer. The Attention layer completes the dynamic weight allocation of the feature output, and the output layer outputs the feature learning results during the model training process, thus completing the basic training of the model.

[0056] It should be noted that the initial values ​​of the core parameters of each layer are set as follows: the number of neurons in the shared LSTM layer is 64-128, the number of neurons in the task-specific LSTM layer is 32-64, and there is 1 neuron each for electric, cold, and hot loads; the inactivation rate of the Dropout layer is 25%; the input layer dimension is the number of features in the load prediction feature set, and the output layer dimension is 3, corresponding to the predicted values ​​of the three types of loads: electric, cold, and hot.

[0057] The model optimization component uses the particle swarm optimization algorithm to iteratively optimize the model parameters after the basic training of the model, and confirms the optimal number of neurons in the shared LSTM layer and the task-specific LSTM layer, as well as the weight allocation rules of the Attention layer, to obtain the MTL-LSTM model.

[0058] It should be noted that the iteration stopping condition for model training can be that the validation set loss value does not decrease for 5 consecutive iterations, or the maximum number of iterations of 100 is reached; the loss function is the mean squared error (MSE), the optimizer can be the Adam optimizer, the initial learning rate is 0.001, and the batch size is 32; the common patterns extracted by the shared LSTM layer include daily and weekly load fluctuations, and the task-specific LSTM layer specifically learns the seasonal characteristics of a single load, such as high heat load in winter and low heat load in summer, and vice versa for cold load.

[0059] The optimization parameters for the particle swarm optimization algorithm can be: 64-128 neurons in the shared LSTM layer, 32-64 neurons in the task-specific LSTM layer, and 0-1 weight coefficients in the Attention layer. The optimization objective is to minimize the mean squared error (MSE) of the validation set. The algorithm iterates 50 times, with 30 particles, and the inertia weight decreases linearly from 0.9 to 0.4. After optimization, the optimal number of neurons in the shared LSTM layer is 96, and the optimal number of neurons in the task-specific LSTM layer is 48, ensuring a balance between model accuracy and computational efficiency.

[0060] The pass / fail criterion for model performance validation can be the coefficient of determination (R²). 2 The average absolute error of electrical load (MAE) must be ≥0.95, and the average absolute error of cooling load (MAE) must be ≤5kW, the average absolute error of heating load (MAE) must be ≤3kW, and the average absolute error of heating load (MAE) must be ≤4kW. If the standard is not met, the model architecture must be adjusted and the model must be retrained.

[0061] In one embodiment, the load forecasting unit includes: a real-time data acquisition element, a real-time data processing element, a total load forecasting element, and a sub-item load forecasting element.

[0062] It should be noted that, optionally, the real-time forecast time granularity is once per hour, and each forecast predicts the load value for the next 24 hours to meet the daily energy dispatching needs of oil and gas field stations; the forecast results need to be pushed to the intelligent control module in real time for power distribution and heat pump start-up and shutdown control.

[0063] The real-time data acquisition element is used to collect real-time operating data of current electricity, cooling, and heating loads at oil and gas field stations, as well as meteorological data such as regional ambient temperature, solar radiation, and air humidity, to form a real-time raw dataset.

[0064] It should be noted that the sampling frequency of real-time data acquisition is consistent with that of historical data, and data transmission can use the ESP8266 WiFi module with a transmission delay of no more than 5 seconds. If data interruption occurs during the acquisition process, the average data of the previous 10 minutes can be used as a temporary replacement, and a fault alarm can be triggered at the same time.

[0065] The real-time data processing element is used to preprocess the real-time raw dataset, screen load participation factors, perform comprehensive correlation analysis and feature classification screening to obtain a real-time load prediction feature set.

[0066] It should be noted that the preprocessing and feature selection methods for real-time data are completely consistent with those for historical data; standardization and normalization are performed using the x-value of the historical dataset. min x max The parameters μ and σ are not recalculated to ensure a uniform data distribution and avoid model prediction bias caused by differences in processing rules.

[0067] The total load forecasting element is used to input the real-time load forecasting feature set into the MTL-LSTM model, output it with different intensities, and then multiply it by different weights and add them together to obtain the total load forecast value.

[0068] It should be noted that the model outputs of different intensities correspond to the prediction results of strong, medium and weak correlation features. The weights of strong correlation features account for 60% and medium correlation features account for 40%. The weights are based on the optimal weight reorganization after optimization by the particle swarm algorithm, and strictly satisfy the constraint that the sum of the weights is 1. The total load prediction value is the superposition of the three types of load prediction values, that is, total load = electric load prediction value + cooling load prediction value + heating load prediction value.

[0069] The sub-item load forecasting element is used to multiply the correlation degree of each sub-item by the total load forecasting value based on the load forecasting feature set to obtain the predicted values ​​of electricity, cooling and heating loads.

[0070] It should be noted that "each item" refers to the load participation factor of electricity, cooling, and heating loads. This LPF value is calculated based on real-time characteristics. The final single load forecast value = total load forecast value × corresponding LPF value, ensuring that the ratio of single load to total load conforms to the actual energy consumption pattern. The forecast results are stored and pushed in the format of "timestamp - electricity load forecast value (kW) - cooling load forecast value (kW) - heating load forecast value (kW)".

[0071] Understandably, the valve on / off and pump start / stop of each subsystem are uniformly controlled by the intelligent control module, which switches the operating mode based on the predicted cooling / heating load values: during peak heating load, i.e., when the predicted value is ≥ 70% of the standard green electricity real-time conversion amount, the ground source heat pump and air source heat pump operate in parallel, and low-grade heat energy is simultaneously connected to the hot water storage tank; during peak cooling load, the medium flow direction of the heat pump evaporator and condenser is switched to achieve the cooling function; during low load periods, i.e., when the predicted value is < 30% of the standard green electricity real-time conversion amount, a single heat pump or direct collector supply mode is adopted to reduce energy consumption; the effective volume of the hot water storage tank is configured according to 30% of the station's average daily heat load to ensure that enough low-grade heat energy can be stored for the heat pump to use; The data acquisition unit includes sensors such as a temperature sensor, a flow sensor, an energy sensor, and a lithium battery SOC sensor. The sampling frequency of the data acquisition unit is consistent with the load forecasting method. The triggering condition for the energy control unit to regulate energy transmission is based on the difference between the real-time wind and solar power generation and the load forecast value. The triggering conditions are: when the difference is ≥10% of the rated load, the energy storage charging is triggered; when the difference is ≤-10% of the rated load, the energy storage discharging is triggered; and when the difference is ≤-30% of the rated load, the multi-source heat pump is triggered to start. like Figure 8 As shown, the intelligent control module can use an STM32F103C8T6 microcontroller as the main controller, supporting multiple sleep modes and a standby current ≤1μA, meeting the requirements for long-term continuous operation of oil and gas field stations; the human-machine interaction layer includes an OLED display panel, an LD3320 voice recognition module, and an ESP8266 WiFi module. The OLED display panel displays real-time wind and solar power generation, lithium battery SOC value and temperature, predicted and actual values ​​of electrical / cooling / heating loads, and the operating status of multi-source heat pumps; the keywords for voice control can include "start heating", "stop cooling", and "adjust temperature XX℃", with a recognition accuracy of ≥95%; the APP remote control supports equipment start / stop, parameter setting, and fault alarm push, with a data update frequency of 1 minute / time.

[0072] It should be noted that the control strategy of the power control unit may include: 1. Green electricity priority supply strategy: When the real-time wind and solar power generation is greater than or equal to the predicted power load for the corresponding period, the green electricity generated by wind and solar power is prioritized for direct supply to the power consumption terminals of the station, and the surplus electricity is used to charge the lithium battery energy storage unit; if the lithium battery SOC reaches the full charge threshold and there is still surplus electricity, combined with the predicted cold / heat load, the solar collector is triggered to supply directly or the multi-source heat pump is operated at low load to convert the surplus green electricity into cold / heat energy and store it in the load-side water tank.

[0073] 2. Energy storage tiered replenishment strategy: When the real-time wind and solar power generation is less than the predicted power load for the corresponding period, control the lithium battery energy storage unit to release stored power to replenish the power supply link until the lithium battery SOC drops to the preset lower limit threshold; if the power load demand still cannot be met after replenishment, restrict the power consumption of non-core auxiliary loads according to the energy consumption priority to ensure the power supply of core process loads.

[0074] 3. Multi-source heat pump coordinated control strategy: Based on the predicted cooling / heating load values, the operation mode of the multi-source heat pump is matched. During the peak heat load, the ground source heat pump and the air source heat pump are started to operate in parallel, and the low-grade heat energy of the hot water storage tank is connected to improve the heating efficiency. When there is a demand for cooling load, the heat pump cooling mode is switched. During the low load period, the single heat pump or the direct supply mode of the collector is used to achieve the matching of energy consumption and energy supply demand.

[0075] 4. Energy load priority strategy: The station's energy consumption is divided into core process loads (wellhead heating, crude oil storage tank insulation, etc.) and auxiliary loads (office building air conditioning, etc.). When energy supply is insufficient, priority is given to ensuring the cooling, heating and electricity needs of core process loads, and the energy consumption of auxiliary loads is dynamically adjusted.

[0076] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the appended claims.

Claims

1. A multi-energy complementary clean energy supply system for oil and gas field stations based on load forecasting, characterized in that, include: Station energy terminals are used to cool / heat oil and gas field stations using electrical energy; The multi-energy complementary power supply module is used to convert solar and wind energy into standard green electricity and then directly supply it to the station's energy consumption terminal. When the real-time conversion amount of standard green electricity is greater than the real-time power load of the station's energy consumption terminal, the excess standard green electricity is stored in the lithium battery. The multi-source heat pump coupling process module includes a ground source heat pump unit, an air source heat pump unit, a heat collection circulation unit, and a load-side water tank circulation unit. It collects low-grade heat energy from the oil and gas field station through the ground source heat pump unit, air source heat pump unit, and heat collection circulation unit. When powered by a lithium battery, it converts the low-grade heat energy into high-grade cold / heat energy through a reverse Carnot cycle. The cold / heat medium is then transported to the oil and gas field station for cooling / heating via the load-side water tank circulation unit. The low-grade heat energy includes ground source heat energy, air source heat energy, and water source heat energy. The intelligent control module includes a data acquisition unit, a data filtering unit, a model building unit, a load forecasting unit, and an energy control unit. The data acquisition unit is used to collect historical energy consumption data of oil and gas field stations and corresponding historical regional meteorological data, and to perform data cleaning, standardization, normalization, and outlier correction to generate a high-quality basic dataset. The data filtering unit is used to perform load participation factor filtering, correlation analysis and feature classification filtering on the high-quality basic dataset to construct a load prediction feature set; The model building unit is used to build a multi-task learning prediction model based on a long short-term memory network, and to train the model using the load prediction feature set to obtain an MTL-LSTM model. The load forecasting unit is used to collect energy consumption data and regional meteorological data of oil and gas field stations in real time. After data preprocessing and feature screening, the data is input into the MTL-LSTM model to obtain preliminary forecast values. After weight optimization and linear weighted fusion, the predicted values ​​of electricity, cooling and heating loads of the station's energy consumption terminals are obtained. The power control unit is used to control the lithium battery to supply power to the multi-source heat pump coupling process module and the station energy terminal based on the predicted values ​​of the electric, cooling, and heating loads when the predicted values ​​of the electric, cooling, and heating loads are greater than the standard green energy real-time conversion amount, so as to realize the supplementary cooling / heating of the oil and gas field station by the multi-source heat pump coupling process module.

2. The oil and gas field station multi-energy complementary clean energy supply system based on load forecasting according to claim 1, characterized in that, The multi-energy complementary power supply module includes: a photovoltaic power generation unit, a wind power generation unit, a voltage regulator and boost circuit, a lithium battery energy storage unit, and an energy detection unit; The power output terminals of the photovoltaic power generation unit and the wind power generation unit are connected to the input terminal of the voltage regulator and boost circuit. The output terminals of the voltage regulator and boost circuit are respectively connected to the station's energy consumption terminal and the charging terminal of the lithium battery energy storage unit; The power detection unit is connected to the photovoltaic power generation unit, the wind power generation unit and the station energy terminal respectively, and is used to detect the wind power generation, photovoltaic power generation and the energy load of the station energy terminal in real time. The surface of the lithium battery energy storage unit is covered with graphene-enhanced phase change thermal storage material, and the discharge end is connected to the power input end of the station energy terminal and the multi-source heat pump coupling process module, respectively.

3. The oil and gas field station multi-energy complementary clean energy supply system based on load forecasting according to claim 1, characterized in that, The solar collector circulation unit includes a PVT collector, a collector circulation pump, and a hot water storage tank. The PVT collector is connected to the hot water storage tank through the collector circulation pump to form a solar solar collector loop, which is used to collect low-grade solar thermal energy and store it in the hot water storage tank. The hot water storage tank is connected to the ground source heat pump unit and the air source heat pump unit respectively, and is used to transfer the stored low-grade thermal energy to the ground source heat pump unit and the air source heat pump unit. The ground source heat pump unit includes a first compressor, a first evaporator, a first condenser, and a first throttling valve, which are connected in sequence to form a first closed refrigerant circulation loop; the first evaporator is connected to the hot water storage tank and is used to collect low-grade heat energy from the ground source and low-grade heat energy in the hot water storage tank; the first condenser is connected to the load-side water tank circulation unit and is used to output high-grade heat energy / cold energy after heat pump circulation conversion. The air source heat pump unit includes a second compressor, a second evaporator, a second condenser, and a second throttling valve, which are connected in sequence to form a second closed refrigerant circulation loop; the second evaporator is connected to the hot water storage tank and is used to collect low-grade heat energy from the air source and low-grade heat energy from the hot water storage tank; the second condenser is connected to the load-side water tank circulation unit and is used to output high-grade heat energy / cold energy after heat pump circulation conversion. The load-side water tank circulation unit includes a load-side water tank, a load-side water tank circulation pump, and a terminal circulation pump. The load-side water tank is connected to the condenser outlets of the ground source heat pump unit and the air source heat pump unit via the load-side water tank circulation pump, respectively, for receiving and storing high-grade heat / cold energy output by the ground source heat pump unit and the air source heat pump unit. The outlet of the load-side water tank is connected to the energy-consuming terminal of the oil and gas field station via the terminal circulation pump, for transporting cold / heat medium to the energy-consuming terminal to achieve energy supply.

4. The oil and gas field station multi-energy complementary clean energy supply system based on load forecasting according to claim 3, characterized in that, The energy-consuming terminal is a finned heat exchanger, used for cooling / heating through the cold / hot medium delivered inside.

5. The oil and gas field station multi-energy complementary clean energy supply system based on load forecasting according to claim 1, characterized in that, The data acquisition unit includes: a historical data acquisition element, a historical data cleaning element, and a historical data processing element; The historical data acquisition element is used to collect historical data on the operation of electricity, cooling, and heating loads of oil and gas field stations in a time series manner, and simultaneously collect the corresponding historical station area ambient temperature, solar radiation and air humidity to form the original historical dataset. The historical data cleaning component is used to initially screen out abnormal data that exceeds the reasonable value range using the statistical threshold method, and then perform secondary verification using the time series trend method and the correlation method to distinguish between single-point anomalies and continuous short-term anomalies. The abnormal values ​​are then corrected using the adjacent time effective value replacement method and the same period and same working condition data replacement method, respectively, thus completing the data cleaning. Historical data processing components are used to standardize and normalize the cleaned dataset to obtain a high-quality basic dataset.

6. The oil and gas field station multi-energy complementary clean energy supply system based on load forecasting according to claim 1, characterized in that, The data filtering unit includes: a factor filtering element, a correlation analysis element, and a feature filtering element; The factor screening element is used to screen the load participation factors of the high-quality basic dataset. The screening rule for the load participation factors is: to determine whether the obtained load participation factors can effectively quantify the correlation between the sub-items of the load and the total load, and whether the contribution of the core load reaches the conventional energy consumption level of the oil and gas field station. The correlation analysis element is used to calculate the correlation between electrical, cooling, and heating load characteristics, as well as between each load characteristic and the meteorological characteristics of ambient temperature, solar radiation, and air humidity, after the load participation factors are screened, using the grey relational analysis method, and to obtain the correlation values ​​between each characteristic. The feature filtering element is used to perform feature classification filtering based on the correlation degree value to obtain a load prediction feature set; wherein, features with correlation degree > 0.6 are defined as strongly correlated core features, features with correlation degree 0.3 ≤ correlation degree ≤ 0.6 are defined as moderately correlated core features, and features with correlation degree < 0.3 are defined as weakly correlated redundant features, thus obtaining a load prediction feature set.

7. The oil and gas field station multi-energy complementary clean energy supply system based on load forecasting according to claim 6, characterized in that, The formula for calculating the load participation factor is as follows: in, For the first Participating factors of class load, Number of load types For the first , For the first The correlation between load type and total load; The formula for calculating the degree of correlation is: in, To quantify the correlation of individual data points, As a reference sequence, For comparing sequences, The resolution coefficient, This represents the global maximum value of the differences between all comparison sequences and the reference sequence at each time step. This represents the global minimum of the differences between all comparison sequences and the reference sequence at each time step. The formula for calculating the correlation degree is: in, For relevance, is the length of the time series, and k is the time index of the time series.

8. The oil and gas field station multi-energy complementary clean energy supply system based on load forecasting according to claim 1, characterized in that, The model building unit includes: architecture building elements, model training elements, and model optimization elements; Architecture building blocks are used to construct a multi-task learning and prediction model architecture based on a long short-term memory network; the model includes a shared LSTM layer, a task-specific LSTM layer, an attention layer, a dropout layer, and an output layer; The model training element is used to input the load prediction feature set into the MTL-LSTM model, extract the temporal common patterns of features through the shared LSTM layer, and then learn the individual patterns of single loads through the task-specific LSTM layer. The Attention layer completes the dynamic weight allocation of the feature output, and the output layer outputs the feature learning results during the model training process, thus completing the basic training of the model. The model optimization component uses the particle swarm optimization algorithm to iteratively optimize the model parameters after the basic training of the model, and confirms the optimal number of neurons in the shared LSTM layer and the task-specific LSTM layer, as well as the weight allocation rules of the Attention layer, to obtain the MTL-LSTM model.

9. The oil and gas field station multi-energy complementary clean energy supply system based on load forecasting according to claim 1, characterized in that, The load forecasting unit includes: a real-time data acquisition element, a real-time data processing element, a total load forecasting element, and a sub-item load forecasting element; The real-time data acquisition element is used to collect real-time operating data of current electricity, cooling and heating loads at oil and gas field stations, as well as meteorological data such as regional ambient temperature, solar radiation, and air humidity, to form a real-time raw dataset. The real-time data processing element is used to preprocess the real-time raw dataset, screen load participation factors, perform comprehensive correlation analysis and feature hierarchical screening to obtain a real-time load prediction feature set. The total load forecasting element is used to input the real-time load forecasting feature set into the MTL-LSTM model, output it with different intensities, and then multiply it by different weights and sum them to obtain the total load forecast value. The sub-item load forecasting element is used to multiply the correlation degree of each sub-item by the total load forecasting value based on the load forecasting feature set to obtain the predicted values ​​of electricity, cooling and heating loads.

10. The oil and gas field station multi-energy complementary clean energy supply system based on load forecasting according to claim 1, characterized in that, The energy consumption data includes electrical load operation data, cooling load operation data, and heating load operation data, and the regional meteorological data includes ambient temperature, solar radiation, and air humidity.

Images