New energy power generation prediction method based on weather data

Abstract

The invention discloses a new energy power generation prediction method based on weather data, and the method comprises the following steps: a, carrying out the characteristic analysis of a raw material, collecting the physical characteristic parameters of a biomass raw material, including the heat value of a dry basis, the moisture content of a wet basis and the mass, and generating an initial power generation prediction through a thermodynamic model; b, a distributed temperature and humidity sensor network is deployed in a raw material storage area, temperature, humidity and rainfall data are collected in real time, and the raw material mildew rate and moisture content change are dynamically predicted in combination with environmental parameters; through multi-dimensional data perception, multi-physical field coupling modeling and an intelligent decision engine, dynamic prediction of the raw material quality and the transportation efficiency in biomass energy power generation is realized, the power generation capacity prediction error is reduced, meanwhile, the raw material loss is reduced, and a key technical support is provided for large-scale efficient utilization of biomass energy power generation in new energy power generation.

Description

Technical Field

[0001] The present invention relates to the field of new energy power generation, and specifically relates to a method for predicting new energy power generation based on weather data. Background Art

[0002] With the global energy structure transforming towards low-carbon, new energy power generation (such as wind energy, solar energy, biomass energy, etc.) has become the core path to achieve the dual-carbon goal. Compared with traditional fossil energy, new energy has advantages such as being renewable and having low pollution. However, its power generation efficiency is significantly affected by environmental factors, and it is necessary to improve the prediction and control capabilities through intelligent technologies to ensure the stability of the power grid.

[0003] In the new energy system, biomass power generation by burning agricultural and forestry waste (such as straw, wood chips) or energy crops has the following irreplaceable features: Abundant resources: Approximately 14 billion tons of biomass waste are generated globally every year, with a utilization rate of less than 10%; Carbon balance characteristics: The carbon dioxide released by combustion can be absorbed by plant photosynthesis, achieving nearly zero carbon emissions; Peak shaving ability: It is not restricted by sunlight and wind speed, and can flexibly adjust the power generation load to make up for the intermittent defects of wind and solar power generation.

[0004] However, biomass power generation faces core challenges: Fluctuations in raw material supply: The moisture content and calorific value of biomass raw materials are significantly affected by weather (rainfall, humidity, temperature) and transportation efficiency (road congestion, vehicle breakdown); Loss of combustion efficiency: Excessive moisture content or mildew will reduce the calorific value utilization rate, resulting in a high deviation in power generation prediction; Extensive management: In existing technologies, it relies on manual experience and lacks the ability for dynamic monitoring and optimization of the entire chain of storage, transportation, and combustion; Currently, two main methods are used for predicting biomass power generation: Static model based on raw material characteristics: Only calculates the theoretical power generation based on the dry basis calorific value and mass measured in the laboratory, ignoring the dynamic losses in the storage and transportation links; Single environmental factor correction model: For example, only adjusts the transportation time through rainfall, without integrating the coupling effects of multiple factors such as snowfall, freezing, and air humidity.

[0005] The above methods have the following problems: Low prediction accuracy: It does not quantify the real-time impact of weather on raw material quality (such as moisture content, mildew rate), resulting in a large deviation in power generation prediction; Response lag: It relies on manual inspections and manual data entry, and cannot dynamically adapt to sudden weather changes or transportation interruptions; Local optimization defect: Only optimize a single link (such as storage or transportation), lacking full-chain collaborative control. Therefore, a method for predicting new energy power generation based on weather data is proposed. Summary of the Invention

[0006] The technical problem to be solved by the present invention is: how to solve the problems of low prediction accuracy and response lag in the existing methods for predicting new energy power generation, and a method for predicting new energy power generation based on weather data is provided.

[0007] The present invention solves the above technical problems through the following technical solutions. The present invention includes the following steps: Step a: Conduct raw material characteristic analysis, collect physical characteristic parameters of biomass raw materials, including dry basis calorific value, wet basis moisture content and mass, and generate an initial power generation prediction through a thermodynamic model. Step b: Deploy a distributed temperature and humidity sensor network in the raw material storage area, collect temperature, humidity and rainfall data in real time, and dynamically predict the mildew rate and moisture content change of the raw materials in combination with environmental parameters. Step c: Conduct transportation dynamic modeling, integrate meteorological satellites, traffic management platforms and on-board diagnostic systems, obtain weather data (including rainfall intensity, snowfall, wind speed, environmental temperature), road grades, real-time traffic flow and vehicle mechanical status of the transportation route in real time, and construct a transportation efficiency prediction model. Step d: Fuse the prediction results of steps a to c through an adaptive weight allocation algorithm to generate a final power generation prediction value. Step e: Conduct risk closed-loop control. When the predicted moisture content exceeds the preset safety threshold or the comprehensive risk value of the transportation route reaches the warning level, automatically trigger the storage area protection device, dynamic optimization of the transportation route and vehicle emergency control protocol.

[0008] Furthermore, in step a, the effective calorific value is calculated through a thermodynamic model, and the initial power generation prediction is generated in combination with the raw material mass and power generation efficiency. The specific process is as follows: ; Where: is the dry basis calorific value of the raw material, is the wet basis moisture content (value range 0-1), is the latent heat of vaporization of water; At the same time, in combination with the power generation efficiency and the raw material mass m, the theoretical power generation is predicted through the formula: ; Calculate the theoretical power generation ; The power generation efficiency Is dynamically calibrated through historical operation data.

[0009] Furthermore, the process of predicting the mildew rate of raw materials is as follows: Calculate the mildew reaction rate through the Arrhenius equation, and the specific formula is: ; In the formula, e is the natural constant (the base of the natural exponential function), used to describe the exponential growth or decay relationship, T is the real-time temperature in the storage area, A and Ea are the kinetic parameters of the mildew reaction calibrated through the laboratory respectively, and R is the ideal gas constant; Use the three-dimensional diffusion equation to simulate the humidity distribution inside the raw material pile, and the control equation is: ; Among them, S(x, y, z) represents the infiltration rate of rainfall or snowmelt at the spatial coordinates (x, y, z) of the raw material pile, and the spatial distribution of precipitation is monitored by a millimeter-wave radar, is the moisture content, D is the humidity diffusion coefficient, is the Laplace operator of humidity (the second-order derivative in three-dimensional space), representing the spatial distribution gradient of humidity, and t is the time variable.

[0010] Furthermore, the dynamic correction method of the humidity diffusion coefficient D is: Nonlinearly adjust the diffusion coefficient according to the real-time humidity H and the snowfall amount S, and the specific formula is as follows: ; In the formula, D0 is the basic diffusion coefficient, the benchmark value of the diffusion ability in a dry environment, H critica is the preset humidity critical threshold, and when the real-time humidity exceeds this threshold, the diffusion coefficient compensation mechanism is activated; H scale is the humidity influence scale parameter, which controls the sensitivity of the diffusion coefficient to changes in humidity; H is the real-time humidity, obtained by the humidity sensor in the storage area; The parameter α s is dynamically calibrated according to the relationship between the snow quality density and the porosity of the raw materials.

[0011] Furthermore, in the process of dynamic modeling of transportation, the prediction of the delay time is carried out, and the specific prediction process is as follows: Calculate the transportation delay time based on multi-source data fusion: ; In the formula, α is the rainfall delay coefficient, which reflects the influence degree of rainfall on the vehicle speed; R trans and S are the rainfall intensity and the snowfall amount respectively, obtained through the fusion of in-vehicle weather station and satellite data; d is the transportation distance, T freezeis the freezing temperature threshold, T env is the ambient temperature, measured by the on-vehicle temperature and humidity sensor; v snow is the vehicle speed attenuation coefficient on the ice and snow road surface, reflecting the influence of temperature on the vehicle speed on the ice and snow road surface; α s is the snowfall delay coefficient; During the transportation dynamic modeling process, the prediction of the increment of transportation moisture content is also carried out; The road grades are divided into three levels: expressway, national highway, and provincial highway, corresponding to different reference vehicle speeds v base , and the vehicle speed is corrected twice in combination with the real-time traffic congestion index TI.

[0012] Furthermore, the specific process of the prediction of the increment of transportation moisture content is as follows: a multi-factor coupling calculation model of rainwater penetration, snowmelt moisture absorption, and air humidity diffusion: ; where H air is the real-time air humidity during transportation, sampled by the on-vehicle temperature and humidity sensor every 5 minutes; The parameter γ is the snowmelt moisture absorption coefficient, which is positively correlated with the raw material surface area / volume ratio, and the geometric shape of the raw material is measured in real time by a multispectral imager; β is the rainwater penetration coefficient, which is related to the tightness of the tarpaulin of the transportation vehicle; km is the snow layer melting rate constant, which is positively correlated with temperature and wind speed.

[0013] δ is the air humidity diffusion coefficient, characterizing the moisture absorption ability of the raw material in a high-humidity environment; H air is the air humidity during transportation, sampled every 5 minutes.

[0014] H eq is the equilibrium humidity of the raw material, obtained by measuring the hygroscopic isotherm in the laboratory; Δt is the transportation delay time, and t is the snow layer exposure time, that is, the cumulative time when the raw material surface is covered by snow.

[0015] Furthermore, the implementation process of the adaptive weight allocation algorithm is as follows: Based on the historical error MAE of each stage prediction model i calculate the dynamic weight: ; where α is the weight allocation sensitivity coefficient, dynamically adjusted by the LSTM network according to the weather stability index, used to control the weight allocation difference degree (default value 1.0); The weight allocation result is updated once every preset time period, and the historical weight changes are recorded through the blockchain evidence storage module. Let λi be the weight of the i-th stage. For i = 1, 2, 3 corresponding to the raw material analysis, storage, and transportation stages, it satisfies λ1 + λ2 + λ3 = 1.

[0016] MAE i is the historical mean absolute error of the prediction model for the i-th stage, calculated from the data of the past 30 days.

[0017] Furthermore, the content of the risk closed-loop control also includes the dehumidification control of the transport vehicle. The dehumidification control process of the transport vehicle is as follows: A humidity sensor is installed on the transport vehicle. When the humidity sensor monitors that the humidity H of the in-vehicle raw material is greater than H threshold , the dehumidification unit is started and the desiccant dosage is controlled. H threshold is the preset humidity critical value. The process of obtaining the desiccant dosage is as follows: ; where k d is the desiccant dosage efficiency coefficient, determined by the desiccant type and the dehumidifier power. H target is the target humidity, that is, the preset safety threshold. V storage is the volume of the storage area, regularly calibrated by 3D point cloud scanning.

[0018] Furthermore, the method for predicting new energy power generation also includes moisture content detection by a multispectral imager. The specific process is as follows: Use a multispectral imager to collect the reflection spectrum of the raw material surface at characteristic wavelengths; Invert the moisture content through a non-linear regression model: ; where to are the intensities of the reflected light at characteristic wavelengths. k1, k2, and k3 are calibration coefficients, obtained by regression fitting of laboratory standard samples and recalibrated every quarter with laboratory standard samples.

[0019] The present invention has the following advantages compared with the prior art: The method for predicting new energy power generation based on weather data overcomes the limitations of traditional single-model prediction by integrating multi-dimensional data of raw material characteristics, storage environment, and transportation dynamics, significantly improving the accuracy of biomass combustion power generation prediction, providing a reliable basis for power grid dispatching, covering the entire process from raw material storage, transportation to combustion power generation, real-time monitoring of moisture content changes and mildew risks, ensuring controllable raw material quality, reducing power generation efficiency losses caused by raw material deterioration, dynamically responding to weather changes such as rainfall, snowfall, freezing, and high humidity, quantifying the impact of weather on storage and transportation links, optimizing response strategies, reducing the risk of power generation interruption caused by extreme weather, combining real-time traffic conditions and road grade differences, dynamically adjusting transportation routes and vehicle speed predictions, balancing transportation timeliness and fuel consumption costs, and improving logistics efficiency; Visual monitoring of the spatial distribution of moisture content is achieved by using multi-spectral imaging technology, avoiding the damage to raw materials caused by traditional sampling detection, identifying local abnormal areas at the same time, improving detection efficiency, automatically triggering emergency measures such as drying in the storage area and heating protection during transportation, combining with path re-planning algorithms, forming a closed loop of monitoring, early warning, and disposal, reducing operation risks, and full-process automatic data collection and analysis, reducing manual inspection and manual entry errors, reducing operation and maintenance costs, and improving the intelligent level of the biomass power generation system. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 is the overall flowchart of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0021] The embodiments of the present invention will be described in detail below. The embodiments are implemented on the premise of the technical solution of the present invention, and detailed implementation manners and specific operation processes are given. However, the protection scope of the present invention is not limited to the following embodiments.

[0022] As Figure 1 shown, this embodiment provides a technical solution: A method for predicting new energy power generation based on weather data, including the following steps: Step a: Analyze the characteristics of raw materials, collect the physical characteristic parameters of biomass raw materials, including dry basis calorific value, wet basis moisture content, and quality, and generate an initial power generation prediction through a thermodynamic model; Step b: Deploy a distributed temperature and humidity sensor network in the raw material storage area, collect temperature, humidity, and rainfall data in real time, and dynamically predict the mildew rate and moisture content change of raw materials in combination with environmental parameters; Step c: Model the transportation dynamics, integrate meteorological satellites, traffic management platforms, and vehicle diagnostic systems, obtain the weather data (including rainfall intensity, snowfall, wind speed, environmental temperature), road grade, real-time traffic flow, and vehicle mechanical status of the transportation route in real time, and construct a transportation efficiency prediction model; Step d: The prediction results of steps a to c are fused through an adaptive weight allocation algorithm to generate the final predicted power generation value; Step e: Perform risk closed-loop control. When the predicted moisture content exceeds the preset safety threshold or the comprehensive risk value of the transportation path reaches the warning level, the storage area protection device, dynamic optimization of the transportation path, and vehicle emergency control protocol are automatically triggered.

[0023] In step a, the effective calorific value is calculated through a thermodynamic model, and the initial predicted power generation is generated by combining the raw material quality and power generation efficiency. The specific process is as follows: ; Where: is the dry basis calorific value of the raw material, is the wet basis moisture content (value range 0 - 1), is the latent heat of vaporization of water; At the same time, combining the power generation efficiency and the raw material quality m, the theoretical power generation is predicted through the formula: ; The theoretical power generation is calculated; The power generation efficiency is dynamically calibrated through historical operation data; Through the effective calorific value formula, the energy loss caused by water evaporation is accurately deducted, avoiding the overestimation of calorific value caused by ignoring the latent heat of vaporization in the existing technology, ensuring that the power generation prediction is closer to the actual combustion efficiency, and reducing the risk of fuel waste or power supply shortage caused by calorific value deviation.

[0024] Directly related to the wet basis moisture content, the effective calorific value can be dynamically adjusted according to the change of raw material humidity, adapting to the humidity fluctuation caused by weather (such as rainfall, moisture absorption) during storage or transportation. In the rainy season or high humidity areas, the system automatically corrects the prediction value to avoid the power generation plan deviation caused by the temporary moisture of raw materials.

[0025] Combining the power generation efficiency and the raw material quality, quantifying the theoretical power generation, providing a basis for fuel procurement and inventory allocation, helping the operator accurately match the raw material inventory with the power generation demand, and reducing the risk of overstocking or shortage; The biomass power station receives two batches of straw raw materials: Batch A: The dry basis calorific value Q dry is 15 MJ / kg, and the moisture content w is 10%; Batch B: The dry basis calorific value is the same Q dry = 1 MJ / kg, but the moisture content w is 25% (due to rainfall during transportation).

[0026] Defects in the existing technology: If only calculated based on the dry - basis calorific value, the theoretical power generation of the two batches is the same. However, in actual combustion: Batch B consumes a large amount of energy due to water evaporation, and its actual power generation is significantly lower than that of Batch A.

[0027] Calculated by the effective calorific value formula: Batch A: Qeff = 15*(1 - 0.1)-2.26*0.1 = 13.5 - 0.226 = 13.274 MJ / kg; Batch B: Qeff = 15*(1 - 0.25)-2.26*0.25 = 11.25 - 0.565 = 10.685 MJ / kgQ; The prior art will misjudge that the power generation of the two batches is the same; The present invention accurately shows that the power generation of Batch B is lower than that of Batch A, guiding the operator to preferentially use Batch A or perform pre - drying treatment on Batch B, solving the problem of the interference of the humidity change of biomass raw materials on power generation prediction, and providing core technical support for fuel quality grading and combustion efficiency optimization.

[0028] The process of predicting the mildew rate of raw materials is as follows: Calculate the mildew reaction rate through the Arrhenius equation. The specific formula is, , where e is the natural constant (the base of the natural exponential function), used to describe the exponential growth or decay relationship, T is the real - time temperature of the storage area, A and Ea are the mildew reaction kinetic parameters calibrated through the laboratory respectively, and R is the ideal gas constant; Use the three - dimensional diffusion equation to simulate the internal humidity distribution of the raw material pile. The governing equation is: ; Among them, S(x,y,z) represents the infiltration rate of rainfall or snowmelt at the spatial coordinates (x,y,z) of the raw material pile. Monitor the spatial distribution of precipitation through millimeter - wave radar, is the moisture content, D is the humidity diffusion coefficient, is the Laplace operator of humidity (the second - order derivative in three - dimensional space), representing the spatial distribution gradient of humidity, and t is the time variable; By monitoring the storage temperature and environmental humidity, dynamically evaluate the mildew rate, give early warning of the risk of raw material spoilage, avoid calorific value loss caused by mildew, reduce raw material waste, and ensure power generation stability.

[0029] Real - time simulate the internal humidity distribution of the raw material pile, identify local high - humidity areas (such as the bottom water - accumulation area or the surface condensation area), and guide targeted ventilation or drying measures to avoid energy waste caused by traditional treatment methods.

[0030] Through the joint analysis of the mildew rate and humidity diffusion, dynamically adjust the storage conditions (such as the temperature control threshold). In humid climates, the safe storage time of raw materials can be extended, and the supply - chain pressure can be relieved; For example, in a biomass power plant, corn straw is stored during the plum rain season, and the humidity in the storage area remains continuously high.

[0031] Defects of the prior art: Manual inspection can only detect surface mildew and cannot predict the risk of internal corruption; Blindly starting the full - field ventilation and drying increases power consumption, but there are still local areas that mildew due to uneven humidity diffusion.

[0032] Improvements of the present invention: If the system monitors that the storage temperature rises to 30 °C and the humidity reaches 85%, it automatically determines that the mildew risk level is high; An early warning is issued in advance, prompting to preferentially use the raw materials of this batch or to process them immediately.

[0033] Humidity distribution positioning: The three - dimensional model shows that a high - humidity core area (such as humidity of 95%) is formed in the northwest corner of the raw material pile due to rainwater penetration; The system directionally starts the dehumidifier below this area instead of full - field ventilation.

[0034] The dynamic correction method of the humidity diffusion coefficient D is as follows: The diffusion coefficient is non - linearly adjusted according to the real - time humidity H and the snowfall S. The specific formula is as follows: ; In the formula, D0 is the basic diffusion coefficient, the benchmark value of the diffusion ability in a dry environment, H critica is the preset humidity critical threshold. When the real - time humidity exceeds this threshold, the diffusion coefficient compensation mechanism is started; H scale is the humidity influence scale parameter, which controls the sensitivity of the diffusion coefficient to the change of humidity; H is the real - time humidity, which is obtained by the humidity sensor in the storage area; The parameter α s is dynamically calibrated according to the relationship between the snow quality density and the porosity of the raw materials; Dynamically adjust the parameters of the humidity diffusion model, respond to the changes of the storage environment in real - time (such as sudden heavy snow or continuous high humidity), avoid the prediction deviation of the traditional fixed - parameter model, more accurately predict the internal moisture migration path of the raw material pile, and reduce the risk of local mildew or over - drying.

[0035] Identify the area with abnormal humidity diffusion (such as the accelerated surface moisture penetration caused by snow cover), and guide the directional placement of desiccants or the adjustment of ventilation strategies.

[0036] Compensate the diffusion coefficient through the snowfall parameter, quantify the indirect influence of snowmelt on the moisture content of the raw materials, and still maintain the humidity prediction error in heavy snow weather to ensure storage safety.

[0037] During the transportation dynamic modeling process, the delay time is predicted. The specific prediction process is as follows: Calculate the transportation delay time based on multi-source data fusion: ; In the formula, α is the rainfall delay coefficient, which reflects the influence degree of rainfall on vehicle speed; R trans and S are the rainfall intensity and snowfall amount respectively, which are obtained through the fusion of in-vehicle weather stations and satellite data; d is the transportation distance, T freeze is the freezing temperature threshold, T env is the environmental temperature, which is measured by the in-vehicle temperature and humidity sensor; v snow is the vehicle speed attenuation coefficient on the ice and snow road surface, which reflects the influence of temperature on the vehicle speed on the ice and snow road surface; α s is the snowfall delay coefficient; During the transportation dynamic modeling process, the prediction of the increment of transportation moisture content is also carried out; The road grades are divided into three levels: expressway, national highway, and provincial highway, corresponding to different reference vehicle speeds v base , and the vehicle speed is corrected twice in combination with the real-time traffic congestion index TI; Integrate weather factors such as rainfall, snowfall, and freezing, and combine the differences in road types (expressway / national highway / provincial highway) to quantify the combined influence of multiple factors on transportation time, avoiding prediction deviations caused by single factors (such as only considering rainfall), and providing a reliable time window for the power generation plan.

[0038] Fuse traffic congestion data in real time, dynamically adjust the reference vehicle speed, preferentially select the route with the highest comprehensive efficiency, reduce delays caused by sudden congestion or accidents, and improve the on-time rate of raw material supply.

[0039] Through the vehicle speed attenuation model on the ice and snow road surface, predict the long-term impact of low-temperature freezing on transportation efficiency, and plan countermeasures such as installing anti-skid chains and starting heating equipment in advance to ensure transportation safety in extreme weather.

[0040] For example, a biomass power plant needs to transport raw materials from a warehouse 200 kilometers away, and encounters the following situations on the way: Moderate snow, snowfall amount S = 5 cm / h, environmental temperature T env = -3°C, the original plan is to travel on the expressway throughout the journey, but a certain section is congested due to an accident, and the congestion index TI = 80; The existing technology only based on the historical average vehicle speed, assuming that the expressway v base= 100 km / h, the estimated transportation time is 2 hours. Due to the icy road surface, the vehicle speed actually drops to 60 km / h, and there is a 1.5-hour delay in the congested section, so the total time increases to 4.5 hours, resulting in the failure of the power generation raw materials to be delivered on time; The present invention dynamically corrects the vehicle speed. According to the snowfall S = 5 cm / h and the ambient temperature Tenv = -3 °C, the vehicle speed attenuation on the icy road surface is calculated to be v actual = 70 km / h. After detecting congestion, it automatically switches to the parallel national road where there is no congestion, TI = 20, and the reference vehicle speed v base = 80 km / h.

[0041] The distance of the new route increases to 220 kilometers, but the comprehensive vehicle speed increases to v actual = 75 km / h, and the total transportation time is reduced to 2.9 hours, only 0.9 hours later than the original plan.

[0042] The specific process of predicting the increment of the transportation moisture content is as follows: a multi-factor coupling calculation model of rainwater penetration, snowmelt moisture absorption and air humidity diffusion: ; where H air is the real-time air humidity during transportation, which is sampled every 5 minutes by an on-vehicle temperature and humidity sensor; The parameter γ is the snowmelt moisture absorption coefficient, which is positively correlated with the surface area / volume ratio of the raw material, and the geometric shape of the raw material is measured in real time by a multi-spectral imager; β is the rainwater penetration coefficient, which is related to the tightness of the tarpaulin of the transportation vehicle; km is the snow layer melting rate constant, which is positively correlated with temperature and wind speed.

[0043] δ is the air humidity diffusion coefficient, which characterizes the moisture absorption ability of the raw material in a high-humidity environment; H air is the air humidity during transportation, which is sampled every 5 minutes; H eq is the equilibrium humidity of the raw material, which is obtained by measuring the moisture sorption isotherm in the laboratory; Δt is the transportation delay time, and t is the snow layer exposure time, that is, the cumulative time when the surface of the raw material is covered by snow; The above process simultaneously considers the combined effects of rainwater penetration, snowmelt moisture absorption and air humidity diffusion on the moisture content of the raw material, breaks through the limitations of the traditional single-factor model, reduces the moisture content prediction error in complex weather (such as sleet + high humidity), avoids the decrease in combustion efficiency caused by underestimating the moisture content, monitors the snowfall, air humidity and penetration time during transportation in real time, dynamically adjusts the weights of each influencing factor, and continuously corrects the predicted value during long-distance transportation to adapt to sudden weather changes (such as clearing up or sudden rainfall midway).

[0044] By quantifying the contribution ratio of different factors to the moisture content, the protection equipment is started targeted (such as covering the tarpaulin for rain protection first rather than heating and dehumidifying), unnecessary energy consumption is reduced (such as overheating due to misjudging the impact of snow melting), and the transportation cost is lowered; For example, when a transportation fleet transports wood chip raw materials from the mountainous area to the power plant, the journey experiences are as follows: The first 2 hours: moderate rain, rainfall intensity R trans is 10 mm / h; The next 3 hours: snowfall, snow volume S is 8 cm / h, and the air humidity rises to 90%.

[0045] Existing technology: Only calculating according to rainwater penetration, it is predicted that the moisture content increases by Δw = 5%; Actually, due to the moisture absorption of the snow layer melting and high humidity diffusion, the moisture content increases by 12%, resulting in insufficient combustion and a decrease in power generation.

[0046] The present invention dynamically corrects through multiple factors: Rainwater penetration, the penetration contribution in the first 2 hours is +4%; Moisture absorption of snow melting: the penetration contribution of the snow melting in the next 3 hours is +5%; Air humidity diffusion: the continuous moisture absorption contribution in a high humidity environment is +3%; Total increment: Δw = 4% + 5% + 3% = 12%.

[0047] The system identifies snow melting as the main risk (accounting for 42%), and preferentially starts the on-vehicle heating device to accelerate snow melting and evaporation; At the same time, the tarpaulin is unfolded to reduce subsequent rainwater penetration, rather than blindly increasing desiccants.

[0048] The solution of the present invention can reduce the actual moisture content from 12% to 9% (by heating and evaporating part of the snow water), and the loss of power generation efficiency is greatly reduced. Compared with the full protection mode of the existing technology, that is, heating, drying and covering at the same time, it is more energy-saving.

[0049] The implementation process of the adaptive weight allocation algorithm is as follows: Based on the historical error MAE of each stage prediction model i Calculate the dynamic weight: ; where α is the weight allocation sensitivity coefficient, which is dynamically adjusted by the LSTM network according to the weather stability index and is used to control the weight allocation difference degree (default value 1.0); The weight allocation result is updated once every preset time period, and the historical weight changes are recorded through the blockchain evidence storage module; λi is the weight of the i-th stage, i = 1, 2, 3 corresponding to the raw material analysis, storage, and transportation stages, and satisfies λ1 + λ2 + λ3 = 1.

[0050] MAE i is the historical mean absolute error of the i-th stage prediction model, calculated from the data of the past 30 days; The present invention adjusts the weights in real time according to the historical prediction errors of each stage (raw material analysis, storage monitoring, transportation modeling) to ensure that the stages with lower errors have a greater impact on the final prediction result.

[0051] Avoid overall prediction distortion caused by abnormal data in a certain stage (such as sensor failure or sudden weather), and improve the reliability of the prediction result.

[0052] Analyze real-time data (such as sudden weather changes, equipment status fluctuations) through the LSTM network, dynamically update the weight allocation strategy, and automatically reduce the weights of the affected stages in case of extreme weather or equipment anomalies to maintain prediction stability.

[0053] Smooth the weight allocation through an exponential function to avoid the impact of severe fluctuations in the error of a single stage on the result. In the long-term operation, the fluctuation range of the system prediction error is reduced, and the feasibility of the power generation plan is enhanced.

[0054] A biomass power plant has experienced the following situations for three consecutive days: Day 1: Sunny, smooth transportation, stable storage environment; Day 2: Heavy rain caused transportation delay and a sudden increase in the humidity of the storage area; Day 3: Abnormal raw material analysis data due to sensor failure.

[0055] Most of the existing technologies use fixed weight allocation (such as 40% for raw material analysis, 30% for storage, and 30% for transportation); On Day 2, heavy rain caused a sharp increase in the transportation prediction error. However, due to the fixed weight, the prediction deviation of the final power generation reached 20%; On Day 3, the abnormal raw material analysis data directly led to the complete failure of the prediction result.

[0056] The present invention has made dynamic weight adjustment: Day 1: The errors of each stage are balanced, and the weight allocation is 35% for raw materials, 40% for storage, and 25% for transportation; Day 2: The transportation error increased to 3 times the historical level, and the weight automatically decreased to 10%, while the storage weight increased to 50%; Day 3: Abnormal raw material analysis data (error exceeding the limit), the weight decreased to 5%, and the storage and transportation weights were adjusted to 60% and 35% respectively.

[0057] Comparison of prediction results: Heavy rain on Day 2: The prediction deviation in the existing technology is 20%, while the deviation of the present invention is only 8%; Day 3 Sensor Failure: Existing methods predict failures, while the present invention maintains within an acceptable range by reducing the weight of abnormal data.

[0058] The content of the risk closed-loop control also includes the dehumidification control of the transport vehicle. The dehumidification control process of the transport vehicle is as follows: A humidity sensor is installed on the transport vehicle. When the humidity sensor monitors that the humidity H of the in-vehicle raw materials is greater than H threshold (where H threshold is the preset upper humidity limit value), start the dehumidification unit and control the desiccant dosage. The process of obtaining the desiccant dosage is as follows: ; where k d is the desiccant dosage efficiency coefficient, which is determined by the desiccant type and the dehumidifier power, H target is the target humidity, i.e., the preset safety threshold, V storage is the storage area volume, which is calibrated regularly by 3D point cloud scanning; The present invention automatically triggers protection measures by setting humidity and temperature thresholds, realizing closed-loop control from monitoring to early warning to disposal, avoiding manual intervention delay, and significantly reducing the risks of raw material loss and equipment damage.

[0059] Dynamically calculate the desiccant dosage according to the humidity difference (H - H target ) and the storage volume V storage to avoid overdosage or underdosage, reduce desiccant waste, and ensure dehumidification efficiency at the same time.

[0060] Combined with the ambient temperature T env and the wind speed v wind to dynamically adjust the heating power to prevent raw materials from freezing or being over-dried.

[0061] Maintain the stability of the raw material moisture content in a low-temperature environment to ensure combustion efficiency; Due to heavy rain, the humidity in the storage area of a power station suddenly rises to H = 92% (the safety threshold H threshold = 85%), and at the same time, a transport vehicle is driving in a -5°C environment and encounters snowfall.

[0062] Most of the existing technologies rely on manual inspections. It takes 2 hours to discover that the humidity exceeds the standard, and the loss of raw material mildew during this period is too high; The transport vehicle did not start heating in time, resulting in the surface of the raw materials freezing, and the calorific value loss during combustion is low.

[0063] The present invention protects the storage area and starts the dehumidifier within 10 seconds after the humidity exceeds the limit, and calculates the desiccant dosage: mdryer = kd * (92% - 85%) * 500 m 3= 0.1 * 7 * 500 = 350 kg, reducing the humidity to 83% within 3 hours with zero raw material loss.

[0064] The temperature sensor detects T env = -5°C, automatically activating the heating blanket and generating a detour path; The heating power is adjusted according to the load m load = 10 tons and the wind speed v wind = 8 m / s to: P heat = 0.5 * [0 - (-5)] * 10000 * (1 + 0.1 * 8) = 5000 * 1.8 = 9000, maintaining the raw material temperature at 2°C with no ice formation.

[0065] The method for predicting new energy power generation further includes moisture content detection by a multispectral imager, and the specific process is as follows: The multispectral imager is used to collect the reflection spectrum of the raw material surface at characteristic wavelengths; Inverting the moisture content through a non-linear regression model: ; where to are the intensities of the reflected light at characteristic wavelengths, and k1, k2, and k3 are calibration coefficients, which are obtained through regression fitting of laboratory standard samples and recalibrated every quarter using laboratory standard samples; The non-contact and full-field scanning of the raw material moisture content is achieved through multispectral imaging technology, avoiding the physical damage to the raw materials caused by traditional sampling detection. It ensures the integrity of the raw materials, is particularly suitable for the quality control of fragile biomass fuels (such as wood pellets), captures the spatial distribution of the moisture content in the raw material pile or transportation vehicle, identifies local abnormal areas (such as rain penetration points or frozen surfaces), accurately locates the problem areas, avoids blind treatment (such as full-field drying), and significantly reduces energy consumption and processing costs.

[0066] Through the combined analysis of multiple bands (such as near-infrared, short-wave infrared), it adapts to different lighting conditions (day and night, sunny and cloudy) and raw material surface states (dry, wet, frozen).

[0067] It still maintains the detection accuracy in complex environments such as rain, snow, and frost.

[0068] During the winter transportation of corn straw raw materials by a biomass power plant, snowfall is encountered on the way, and some of the raw material surfaces are frozen.

[0069] Most of the existing technologies are manual sampling detections, which can only measure the unfrozen areas and will misjudge the overall moisture content; The raw materials under the actual ice layer have a higher humidity due to the penetration of snowmelt water.

[0070] For the multispectral imaging detection of the present invention, an in-vehicle multispectral camera scans the surface of the raw materials, identifies the icing area through the characteristic wavelength of ice crystals (1450 nm), and combines the near-infrared band (970 nm) to penetrate the ice layer to detect the true moisture content of the raw materials below.

[0071] When the system detects that the local humidity under the ice layer is 28%, the heating power is automatically calculated as follows: P heat =kh*(0℃ - (-5℃))*5000 kg*(1 + 0.1*10 m / s)=7500 W; Directly heat the icing area to melt the ice layer and reduce the humidity within 2 hours; The use of a multispectral imager for moisture content detection is mainly applied in the following two stages: Deploy a multispectral imager in the raw material storage area to perform non-contact and large-scale scanning of the raw material piles, and monitor the spatial distribution changes of the moisture content during the storage process in real time.

[0072] Collect the reflection spectra of the raw material surface at characteristic wavelengths (such as 970 nm, 1450 nm, etc.) through a multispectral camera; Calculate the moisture content using the inversion formula: , scan the entire warehouse once every 2 hours, and increase the frequency to once every 30 minutes during extreme weather, so as to monitor the areas with abnormal moisture content caused by mildew or moisture absorption, and guide the directional placement of desiccants and ventilation strategies; Dynamic transportation monitoring: Install an in-vehicle multispectral imaging device on the transport vehicle to monitor the dynamic changes of the moisture content of the raw materials during transportation in real time.

[0073] Collect the raw material surface reflection data through an in-vehicle spectrometer, combine the GPS positioning and weather data to analyze the influence of rain penetration and snowmelt moisture absorption on the moisture content, sample once every 10 minutes, and increase the frequency to once every 5 minutes during rainfall / snowfall; Thereby, the moisture content increment parameter Δwtrans in the transportation delay model is corrected in real time, and the instruction to start covering the tarpaulin or heating blanket is triggered.

[0074] In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include at least one of such features. In the description of the present invention, "a plurality" means at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0075] In the description of this specification, the descriptions referring to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples", etc. mean that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0076] Although the embodiments of the present invention have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for predicting new energy power generation based on weather data, characterized in that, It includes the following steps: Step a: Conduct raw material characteristic analysis, collect physical characteristic parameters of biomass raw materials, including dry basis calorific value, wet basis moisture content and mass, and generate an initial power generation prediction through a thermodynamic model; Step b: Deploy a distributed temperature and humidity sensor network in the raw material storage area, collect temperature, humidity and rainfall data in real time, and dynamically predict the raw material mildew rate and moisture content change in combination with environmental parameters; Step c: Conduct transportation dynamic modeling, integrate meteorological satellites, traffic management platforms and vehicle diagnostic systems, obtain weather data, road grades, real-time traffic flow and vehicle mechanical status of the transportation route in real time, and build a transportation efficiency prediction model; Step d: Use an adaptive weight allocation algorithm to fuse the prediction results of steps a to c to generate a final power generation prediction value; Step e: Conduct risk closed-loop control. When the predicted moisture content exceeds the preset safety threshold or the comprehensive risk value of the transportation route reaches the warning level, automatically trigger the storage area protection device, dynamic optimization of the transportation route and vehicle emergency control protocol.

2. The method for predicting new energy power generation based on weather data according to claim 1, characterized in that: In step a, the effective calorific value is calculated through a thermodynamic model, and an initial power generation prediction is generated in combination with the raw material quality and power generation efficiency. The specific process is as follows: ； Wherein: is the dry basis calorific value of the raw material, is the moisture content on wet basis, is the latent heat of vaporization of water; Meanwhile, in combination with the power generation efficiency and the raw material quality m, the theoretical power generation is predicted through the formula: ； Calculate the theoretical power generation ; Power generation efficiency Dynamically calibrated through historical operation data.

3. The method for predicting new energy power generation based on weather data according to claim 2, wherein: The process of predicting the raw material mildew rate is as follows: Calculate the mildew reaction rate through the Arrhenius equation. The specific formula is ； Where e is the natural constant used to describe the exponential growth or decay relationship, T is the real-time temperature of the storage area, A and E a are respectively the kinetic parameters of the mildew reaction calibrated through the laboratory, and R is the ideal gas constant; Use the three-dimensional diffusion equation to simulate the humidity distribution inside the raw material pile. The control equation is: ； Among them, S(x, y, z) represents the infiltration rate of rainfall or snowmelt at the spatial coordinates (x, y, z) of the raw material pile. The spatial distribution of precipitation is monitored by a millimeter-wave radar. is the moisture content, and D is the moisture diffusion coefficient. is the Laplacian operator of humidity, representing the spatial distribution gradient of humidity, and t is the time variable.

4. A method for predicting new energy power generation based on weather data according to claim 3, characterized in that: The dynamic correction method of the humidity diffusion coefficient D is: Non-linearly adjust the diffusion coefficient according to the real-time humidity H and snowfall S. The specific formula is as follows: ； Where D0 is the base diffusion coefficient, the reference value of the diffusion ability in a dry environment, and H critica is a preset humidity critical threshold, and when the real-time humidity exceeds this threshold, the diffusion coefficient compensation mechanism is activated; H scale is the humidity influence scale parameter that controls the sensitivity of the diffusion coefficient to humidity changes; H is the real-time humidity, obtained by the humidity sensor in the storage area; Parameter α s Dynamically calibrated according to the relationship between snow density and raw material porosity.

5. A method for predicting new energy power generation based on weather data according to claim 1, characterized in that: During the transportation dynamic modeling process, predict the delay time. The specific prediction process is as follows: Calculate the transportation delay time based on multi-source data fusion: ； In the formula, α is the rainfall delay coefficient, reflecting the influence degree of rainfall on vehicle speed; R trans R and S are the rainfall intensity and snowfall amount respectively, which are obtained by fusing the on-vehicle meteorological station and satellite data; d is the transportation distance, T freeze is the freezing temperature threshold, T env is the ambient temperature, measured by an on-vehicle temperature and humidity sensor; v snow is the vehicle speed attenuation coefficient on ice and snow roads, reflecting the influence of temperature on the vehicle speed on ice and snow roads; α s is the snowfall delay coefficient; During the transportation dynamic modeling process, the transportation moisture content increment prediction is also carried out; The road grades are divided into three levels: expressways, national highways, and provincial highways, corresponding to different reference vehicle speeds v base , and the vehicle speed is corrected twice in combination with the real-time traffic congestion index TI.

6. The method for predicting new energy power generation based on weather data according to claim 5, wherein: The specific process of the transportation moisture content increment prediction is as follows: A multi-factor coupling calculation model of rainwater penetration, snowmelt moisture absorption and air humidity diffusion: ； Among which H air is the real-time air humidity during transportation, sampled every 5 minutes by an on-vehicle temperature and humidity sensor; The parameter γ is the snowmelt moisture absorption coefficient, which is positively correlated with the raw material surface area / volume ratio, and the raw material geometric shape is measured in real time through a multi-spectral imager; β is the rainwater penetration coefficient, which is related to the tightness of the tarpaulin of the transportation vehicle; km is the snow layer melting rate constant, which is positively correlated with temperature and wind speed; δ is the air humidity diffusion coefficient, characterizing the moisture absorption ability of raw materials in a high humidity environment; H air For the air humidity during transportation, samples are taken every 5 minutes; H eq is the equilibrium moisture content of the raw material, obtained by measuring the laboratory moisture sorption isotherm; Δt is the transportation delay time, t is the snow layer exposure time, that is, the cumulative time when the raw material surface is covered by snow; 7. A method for predicting new energy power generation based on weather data according to claim 1, characterized in that: The implementation process of the adaptive weight allocation algorithm is as follows: Historical error MAE of the prediction model for each stage i Calculate the dynamic weight: ； Where α is the weight allocation sensitivity coefficient, which is dynamically adjusted according to the weather stability index through the LSTM network, and is used to control the weight allocation difference degree; The weight allocation result is updated every preset time period, and the historical weight change is recorded through the blockchain evidence storage module; λi is the weight of the i-th stage, i = 1, 2, 3 corresponding to the raw material analysis, storage and transportation stages, and satisfies λ1 + λ2 + λ3 = 1; MAE i It is the historical mean absolute error of the prediction model for the i-th stage, calculated based on the data of the past 30 days.

8. A method for predicting new energy power generation based on weather data according to claim 1, characterized in that: The content of risk closed-loop control also includes the dehumidification control of transportation vehicles. The dehumidification control process of transportation vehicles is as follows: A humidity sensor is installed on the transport vehicle. When the humidity sensor monitors that the humidity H of the raw materials on the vehicle is greater than H threshold , the dehumidification unit is started and the dosage of the desiccant is controlled. H threshold is a preset humidity critical value. The process of obtaining the dosage of the dryer is as follows: ； where k d is the desiccant delivery efficiency coefficient, determined by the desiccant type and the dehumidifier power, H target is the target humidity, i.e., the preset safety threshold, V storage is the storage area volume, which is calibrated regularly by 3D point cloud scanning.

9. A method for predicting new energy power generation based on weather data according to claim 1, characterized in that: The method for predicting new energy power generation further includes detecting the moisture content by a multispectral imager, and the specific process is as follows: using the multispectral imager to collect the reflection spectrum of the raw material surface at the characteristic wavelength; inverting the moisture content through a non-linear regression model: ； Among them to are the characteristic wavelength reflected light intensities, k1, k2 and k3 are calibration coefficients, which are obtained by regression fitting of laboratory standard samples and recalibrated every quarter through laboratory standard samples.

Images