CN-122026329-A - Load prediction method and system based on user order demands

CN122026329ACN 122026329 ACN122026329 ACN 122026329ACN-122026329-A

Abstract

A load prediction method and system based on user order demands relates to the technical field of power load prediction. The method comprises the steps of calculating standard production load of a single product based on historical production logs and historical power load data, determining characteristic operation power values corresponding to each product type according to the standard production load, generating a base line load prediction curve based on planned starting and ending time and the characteristic operation power values in production plan data, combining the base line load prediction curve with the historical power load data to obtain input characteristics, inputting the input characteristics into a time sequence prediction model to obtain an initial load prediction curve, determining efficiency correction factors and environment correction factors based on the historical production logs and the historical power load data, and correcting the initial load prediction curve according to the efficiency correction factors and the environment correction factors to obtain a final load prediction curve. By implementing the technical scheme provided by the application, the accuracy of industrial enterprise power load prediction is improved.

Inventors

PAN YINGCHAO
DUAN XIAOHAN

Assignees

北京如实智慧电力科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260203

Claims (10)

1. A method of load forecasting based on user order demand, the method comprising: Calculating standard production loads of single products produced by a single production line based on the historical production logs and the historical power load data, and calculating standard non-production loads of all production lines in a non-production time period; Calculating the difference value between the standard production load and the standard non-production load to obtain a net increase average load corresponding to the single product, and calculating the net increase average load of a plurality of historical production tasks belonging to the same product type to obtain a characteristic operation power value corresponding to each product type; Acquiring the planned start-stop time of a target production task from the production plan data, and generating a baseline load prediction curve of the target production task according to the planned start-stop time and the characteristic running power value corresponding to the target production task; combining the baseline load prediction curve and the historical power load data to obtain input characteristics, and inputting the input characteristics into a time sequence prediction model to obtain an initial load prediction curve in a prediction period; Extracting reference operation efficiency and accumulated operation duration of production equipment from the historical production log, and calculating the reference operation efficiency based on the accumulated operation duration to obtain an efficiency correction factor; Determining an environmental correction factor based on the historical operating condition temperatures in the historical production log and the historical power load data; And correcting the initial load prediction curve according to the efficiency correction factor and the environment correction factor to obtain a final load prediction curve.
2. The method according to claim 1, wherein the generating a baseline load prediction curve of the target production task according to the planned start-stop time and the characteristic operation power value corresponding to the target production task specifically includes: Determining a target product type from the target production task, and calling the starting transition time and the stopping transition time of the target product type; Dividing the complete cycle of the target production task based on the planned start-stop time, the start-up transition time and the stop transition time to obtain a start-up transition time period, an operation time period and a stop transition time period; Acquiring a planned starting time and a planned ending time from the planned starting and ending time, acquiring a plurality of starting time points based on the planned starting time and the starting transition time, and taking the ratio of the characteristic running power value to the starting transition time as a starting slope; Multiplying the difference value between each starting time point and the planned starting time by the starting slope to obtain a starting load value corresponding to each starting time point, and sequencing a plurality of starting load values in time sequence to obtain a load sequence of the starting transition period; Adding the planned starting time and the starting transition time to obtain the starting time of the operation time period, and taking the difference value between the planned ending time and the shutdown transition time period as the ending time of the operation time period; Determining a plurality of steady-state time points based on the starting time and the ending time, taking the characteristic running power value as a steady-state load value corresponding to each steady-state time point, and sequencing all the steady-state load values in time sequence to obtain a load sequence of the running time period; generating a plurality of downtime points based on the start time of the downtime transition period by taking the difference between the planned end time and the downtime transition period as the start time of the downtime transition period; Dividing the negative value of the characteristic operation power value by the stop transition time length to obtain a stop slope, calculating the difference value between each stop time point and the starting time of the stop transition time period to obtain a time span, multiplying the time span by the stop slope and adding the characteristic operation power value to obtain a stop load value corresponding to each stop time point, and sequencing the stop load values in time sequence to obtain a load sequence of the stop transition time period; And splicing the load sequences of the starting transition period, the running period and the stopping transition period in time sequence to obtain the baseline load prediction curve.
3. The method according to claim 2, wherein said combining said baseline load prediction curve and said historical power load data results in an input signature, comprising in particular: Determining a target timeline based on the time range of the historical power load data and the time range of the baseline load prediction curve, mapping the historical power load data and the baseline load prediction curve onto the target timeline; Selecting a target time point from the target time axis, intercepting a plurality of historical load values in a preset historical window length before the target time point from the historical power load data, and sequencing the plurality of historical load values in time sequence to obtain a historical load characteristic sequence, wherein the target time point is any time point on the target time axis; performing numerical coding on the time attribute information of the target time point to obtain time attribute characteristics; Intercepting a plurality of baseline load predicted values in a preset guide window from the baseline load predicted curve based on the target time point, and sequencing the plurality of baseline load predicted values in time sequence to obtain a baseline guide characteristic sequence; Splicing and combining the historical load characteristic sequence, the time attribute characteristic and the baseline guide characteristic sequence corresponding to the target time point to obtain a high-dimensional characteristic vector; and combining the high-dimensional feature vectors corresponding to all the target time points in the prediction period according to time sequence to obtain the input feature.
4. The method according to claim 1, wherein the calculating the reference operating efficiency based on the accumulated operating time length to obtain an efficiency correction factor specifically includes: Taking the characteristic operation power value as the reference net increase power consumption of the product type in the ideal state of the equipment, retrieving the latest maintenance time point of the equipment from the historical production log, and determining the accumulated operation duration according to the current time point and the latest maintenance time point; when the accumulated running time length is longer than a preset calibration periodic time length, marking the current time point as a history efficiency anchor point; Subtracting the reference non-production load from the actual average load of the historical production task to obtain actual net increase power consumption of the anchor point corresponding to the historical efficiency anchor point; Performing difference calculation on actual net increase power consumption of each corresponding anchor point of two adjacent historical efficiency anchor points in time to obtain power consumption variation, dividing the power consumption variation by operation time between the two historical efficiency anchor points to obtain a segmented power consumption increment gradient in a target operation interval, wherein the target operation interval is an interval corresponding to the two adjacent historical efficiency anchor points in time; sequencing the plurality of segmented power consumption incremental gradients to obtain a historical power consumption incremental gradient sequence, and selecting the last segmented power consumption incremental gradient from the historical power consumption incremental gradient sequence as a prediction reference gradient; calculating the average accumulated operation time length of the target production task, calculating the difference value between the average accumulated operation time length and the target accumulated operation time length when the target accumulated operation time length is recently marked as the historical efficiency anchor point to obtain a predicted interval operation time length, and multiplying the predicted interval operation time length by the predicted reference gradient to obtain a predicted power consumption increment; And adding the predicted power consumption increment and the actual net increase power consumption of the anchor point corresponding to the anchor point of the latest historical efficiency to obtain a predicted net increase average load, and dividing the predicted net increase average load by the reference net increase power consumption to obtain the efficiency correction factor.
5. The method of claim 1, wherein said determining an environmental correction factor based on the historical operating temperature in the historical production log and the historical power load data, comprises: Extracting environmental temperature data during all historical production tasks from the historical production log, and determining a temperature range covering all historical working conditions according to a plurality of environmental temperature data; dividing the temperature range to obtain a plurality of continuous non-overlapping working condition temperature intervals, and selecting the working condition temperature interval with the largest number of historical production tasks as a reference temperature interval; Taking the difference value of the standard production load and the reference non-production load corresponding to each historical production task as actual net increase power consumption, and dividing the actual net increase power consumption by the characteristic operation power value of the historical production task to obtain comprehensive power consumption skewness; Acquiring the equipment accumulated operation time length of each historical production task, and determining a historical equipment efficiency correction factor according to the equipment accumulated operation time length; Taking the ratio of the integrated power consumption bias to the historical equipment efficiency correction factor as an environmental influence coefficient; Binding the environmental influence coefficients with each historical production task, inducing the average working condition temperatures of all the historical production tasks into each working condition temperature interval, and calculating all the environmental influence coefficients contained in each working condition temperature interval to obtain interval average environmental influence coefficients; And carrying out load prediction on the target production task, obtaining a predicted environmental temperature in the prediction period, comparing the predicted environmental temperature with each working condition temperature interval, determining an interval average environmental impact coefficient corresponding to the target historical production task according to a comparison result, and taking the interval environmental impact coefficient as the environmental correction factor.
6. The method according to claim 1, wherein said correcting said initial load prediction curve according to said efficiency correction factor and said environmental correction factor results in a final load prediction curve, comprising: Acquiring a current prediction time point from the prediction period, and acquiring an initial prediction load value corresponding to the current prediction time point from the initial load prediction curve; Calling a baseline load value corresponding to the current prediction time point from the baseline load prediction curve; multiplying the efficiency correction factor by the environment correction factor to obtain a comprehensive correction factor; multiplying the baseline load value by the comprehensive correction coefficient to obtain corrected production net load; subtracting the baseline load value from the initial predicted load value to obtain a non-production-related reference load component, and adding the corrected production net load and the non-production-related reference load component to obtain a final predicted load value corresponding to the current predicted time point; and combining the final predicted load values corresponding to all the predicted time points in time sequence to obtain the final load prediction curve.
7. The method of claim 6, wherein after said correcting said initial load prediction curve based on said efficiency correction factor and said environmental correction factor to obtain a final load prediction curve, said method further comprises: when the target production task has burst faults in the actual execution process, the type and the influence range of the burst faults are obtained, and the scheduled delivery time of the target production task is called from the production schedule data; calculating the interval duration of the scheduled delivery time and the current time, and analyzing the type and the influence range of the sudden fault according to the interval duration to obtain a fault processing grade; matching from a preset emergency scheme library based on the fault handling grade and the type of the sudden fault to obtain a target emergency handling measure; Calculating the expected power load change generated by executing the target emergency treatment measures, and generating an emergency load curve from the fault occurrence time point; and based on the fault occurrence time point, superposing or replacing and correcting the emergency load curve and the final load prediction curve to obtain a real-time corrected emergency load prediction curve.
8. A load prediction system based on user order demands is characterized in that the system comprises an acquisition unit, a calculation unit, a processing unit and a correction unit, The system comprises an acquisition unit, a reference non-production load calculation unit, a production line calculation unit and a production line calculation unit, wherein the acquisition unit acquires production plan data, historical production logs and historical power load data of a target enterprise; the calculation unit calculates the difference value between the standard production load and the standard non-production load to obtain a net increase average load corresponding to the single product, and calculates the net increase average loads of a plurality of historical production tasks belonging to the same product type to obtain characteristic operation power values corresponding to each product type; The processing unit acquires the planned start-stop time of a target production task from the production plan data, generates a baseline load prediction curve of the target production task according to the planned start-stop time and a characteristic operation power value corresponding to the target production task, combines the baseline load prediction curve and the historical power load data to obtain input characteristics, inputs the input characteristics into a time sequence prediction model to obtain an initial load prediction curve in a prediction period, extracts the reference operation efficiency and the accumulated operation duration of production equipment from the historical production log, calculates the reference operation efficiency based on the accumulated operation duration to obtain an efficiency correction factor, and determines an environment correction factor based on the historical working condition temperature and the historical power load data in the historical production log; And the correction unit corrects the initial load prediction curve according to the efficiency correction factor and the environment correction factor to obtain a final load prediction curve.
9. An electronic device comprising a processor, a memory, a user interface, and a network interface, the memory for storing instructions, the user interface and the network interface for communicating with other devices, the processor for executing the instructions stored in the memory to cause the electronic device to perform the method of any of claims 1-7.
10. A computer readable storage medium storing instructions which, when executed, perform the method of any one of claims 1-7.

Description

Load prediction method and system based on user order demands Technical Field The application relates to the technical field of power load prediction, in particular to a load prediction method and system based on user order demands. Background Under the dual drive of the 'two carbon' goal and energy cost optimization, the refined energy management of industrial enterprises is increasingly important. As a core technology of the power load prediction, the prediction precision directly influences the production cost control of enterprises, the power market trading strategy and the stable operation of a power grid, so that the research and application of the high-precision industrial load prediction method has remarkable economic and social values. In the prior art, the mainstream load prediction method generally adopts a time sequence model, such as a long-term and short-term memory network. The method constructs a prediction model by learning inherent statistical rules such as periodicity, trending and the like in historical power load data and possibly combining external variables such as date type, weather and the like, and deduces a future power load curve. However, in the prior art, the prediction object is regarded as a "black box", and is extrapolated mainly depending on the statistical characteristics of the historical load sequence, so that specific production plan information such as future production orders, product types, planned start-stop times and the like cannot be integrated into the prediction model. When the production plan and the history mode of the enterprise deviate greatly, the model which only depends on the history data can not capture the discontinuous structural change, so that the prediction precision is reduced, and the actual requirements of the enterprise on fine production and cost management and control are difficult to meet. Disclosure of Invention The application provides a load prediction method and a system based on user order demands, wherein the method comprehensively considers information of production plans, equipment states and environmental factors, and improves the accuracy of power load prediction of industrial enterprises. In a first aspect, the present application provides a load forecasting method based on user order demand, the method comprising obtaining production planning data, historical production logs, and historical power load data of a target enterprise; the method comprises the steps of calculating standard production load of a single product of a single production line based on historical production logs and historical power load data, calculating reference non-production load of all production lines in a non-production time period, calculating difference values of the standard production load and the reference non-production load to obtain net-increase average load corresponding to the single product, carrying out average calculation on net-increase average loads of a plurality of historical production tasks belonging to the same product type to obtain characteristic operation power values corresponding to each product type, obtaining planned starting and ending time of a target production task from production plan data, generating a base line load prediction curve of the target production task according to the planned starting and ending time and the characteristic operation power values corresponding to the target production task, merging the base line load prediction curve and the historical power load data to obtain input characteristics, inputting the input characteristics into a time sequence prediction model to obtain an initial load prediction curve in a prediction period, extracting reference operation efficiency and accumulated operation time length of production equipment from the historical production logs, calculating an efficiency correction factor based on accumulated operation time length to obtain historical working condition temperature and historical power load data in the historical production log, determining an environment correction factor, and finally carrying out prediction curve according to the efficiency correction factor and the environment correction factor. According to the technical scheme, standard production load of a single product is calculated based on historical production logs and historical power load data, reference non-production load of a production line in a non-production time period is calculated, net increase average load corresponding to the single product is obtained through difference calculation of the standard production load and the reference non-production load, characteristic operation power values corresponding to each product type are obtained through calculation of net increase average loads of a plurality of historical production tasks belonging to the same product type, a baseline load prediction curve is generated according to planned starting and stopping time and the characteristic operat