Hierarchical fine net load prediction and lean distribution network planning system

Abstract

The invention relates to a hierarchical fine net load prediction and lean distribution network planning system, and belongs to the technical field of smart power grids. Comprising a load prediction engine which generates net load curves of a transformer area, a feeder line and a transformer substation according to a spatial hierarchy; the planning decision platform outputs a grid optimization scheme based on the prediction result; the dynamic feedback module is used for collecting operation data in real time to correct the prediction and planning model; according to the invention, the net load prediction error is reduced to below 7%; the utilization rate of distribution network equipment is increased to more than 65%; and the new energy consumption rate is increased by 15-30%. The invention relates to an integrated system integrating multi-level net load prediction and dynamic lean planning.

Description

TECHNICAL FIELD

[0001] The present application relates to a hierarchical fine net load prediction and lean distribution network planning system, belonging to the technical field of smart grid. BACKGROUND

[0002] With the large-scale access of distributed photovoltaic, energy storage and electric vehicles, the load characteristics of distribution network have fundamentally changed, and the traditional planning method faces three major challenges:

[0003] Insufficient load prediction accuracy: existing methods are mostly based on historical total load curve, without distinguishing between basic load and flexible load (such as air conditioning group, charging pile); the reverse power of new energy causes the "net load" to fluctuate sharply, and the error of traditional time series prediction model is more than 15%, which cannot support accurate planning.

[0004] Planning and operation are disconnected: distribution network expansion relies on the maximum load peak value, ignoring the potential of space-time transfer, resulting in low equipment utilization (average <40%); new energy consumption and network loss optimization are not included in the planning stage, and frequent modification is needed in actual operation.

[0005] Data dimension is missing: dynamic factors such as weather, user behavior and electricity price policy are not quantitatively coupled; the data of transformer area, feeder and substation are isolated at different levels, and there is a lack of collaborative analysis mechanism.

[0006] The main defects of existing solutions include:

[0007] Load prediction: single models such as ARIMA and neural network are used, which have poor adaptability to the steep change characteristics of "net load" (total load-new energy output);

[0008] Planning method: static scenario simulation does not consider the demand response potential, resulting in capacitor redundancy or voltage out of limit;

[0009] System architecture: the prediction module and the planning module run independently, and manual data import is needed in between, which is time-consuming and error-prone. SUMMARY

[0010] According to the problems described in the background, the present application solves the problems of:

[0011] It is urgent to develop an integrated system that integrates multi-level net load prediction and dynamic lean planning to solve the planning adaptability problem of new energy distribution network.

[0012] To achieve the above purpose, the present application provides the following technical solutions:

[0013] A hierarchical fine load forecasting and lean distribution network planning system, comprising: a load forecasting engine generating the net load curve of the substation, feeder and substation according to the spatial level; a planning decision platform outputting the network optimization scheme based on the forecasting result; a dynamic feedback module collecting real-time operation data to correct the forecasting and planning model; wherein the load forecasting engine comprises:

[0014] a basic load forecasting unit decomposing rigid electricity load;

[0015] a flexible load response model quantifying the influence of electricity price / temperature on adjustable load;

[0016] a new energy output corrector fusing meteorological satellite data to calibrate photovoltaic / wind power forecasting.

[0017] Preferably, the load forecasting engine performs hierarchical forecasting:

[0018] Step S1: Non-intrusive identification technology is used at the substation level to separate flexible loads such as air conditioners and charging piles;

[0019] Step S2: The feeder level superimposes the substation net load and correlates the topological impedance to calculate line loss;

[0020] Step S3: The substation level aggregates feeder data and introduces transfer load constraints to generate regional net load curve.

[0021] Preferably, the flexible load response model comprises:

[0022] an air conditioner cluster equivalent thermal parameter model calculating power offset under temperature change;

[0023] an electric vehicle charging probability model generating charging spatio-temporal distribution based on user travel big data.

[0024] Preferably, the planning decision platform comprises:

[0025] a capacity optimization module configuring energy storage location and capacity with the goal of reducing net load peak by 20%;

[0026] a network reconstruction algorithm identifying weak links based on power flow entropy index and generating line capacity increase / segmentation scheme;

[0027] an economic evaluator comparing the life cycle cost of the planning scheme with new energy consumption income.

[0028] Preferably, the capacity optimization module performs:

[0029] When the predicted feeder net load is continuously greater than 90% of the limit value, start the energy storage charging instruction;

[0030] When new energy output drops sharply, call energy storage discharge and trigger demand response compensation mechanism.

[0031] Preferably, the dynamic feedback module comprises:

[0032] an error tracking unit, calculating the deviation rate of the actual net load from the predicted value;

[0033] a model self-learning unit, automatically adjusting the load decomposition weight coefficient when the deviation rate is >10%;

[0034] a planning verification interface, feeding the operation voltage out-of-limit and equipment overload events to the planning decision platform.

[0035] Preferably, the system is configured with a multi-scenario simulator to generate an elastic planning scheme under extreme high temperature, new energy off-grid and other scenarios, and output a risk level assessment report.

[0036] Preferably, the basic load prediction unit accesses the high-frequency data of smart meters, and identifies impact loads such as industrial electric arc furnaces and rolling mills through a load characteristic waveform library.

[0037] Preferably, the network frame reconstruction algorithm introduces a dynamic load capacity ratio index, constrains the transformer load rate in the 40%-80% safe interval, and preferentially deploys distributed power access points.

[0038] Preferably, the system is data-interoperable with the power grid dispatching CMS system and marketing management system, realizing a "prediction-planning-operation-verification" closed-loop control.

[0039] The beneficial effects of the present application are:

[0040] The net load prediction error of the present application is reduced to below 7%; the distribution network equipment utilization rate is improved to above 65%; and the new energy consumption rate is increased by 15%-30%. The present application is an integrated system that combines multi-level net load prediction and dynamic lean planning. DETAILED DESCRIPTION

[0041] Example 1

[0042] A hierarchical fine net load prediction and lean distribution network planning system, comprising: a load prediction engine generating net load curves of a substation, a feeder and a transformer station according to spatial levels; a planning decision platform outputting a network frame optimization scheme based on the prediction results; a dynamic feedback module collecting operation data in real time to correct the prediction and planning model; wherein the load prediction engine comprises:

[0043] a basic load prediction unit decomposing rigid electricity consumption load;

[0044] a flexible load response model quantifying the influence of electricity price / temperature on adjustable load;

[0045] a new energy output corrector fusing meteorological satellite data to calibrate photovoltaic / wind power prediction.

[0046] The load prediction engine performs hierarchical prediction:

[0047] Step S1: At the transformer area level, non-intrusive identification technology is used to separate flexible loads such as air conditioners and charging piles.

[0048] Step S2: At the feeder level, superimpose the transformer area net load to calculate line loss in association with the topology impedance.

[0049] Step S3: At the substation level, aggregate feeder data and introduce transfer load constraints to generate regional net load curves.

[0050] The flexible load response model includes:

[0051] Air conditioner cluster equivalent thermal parameter model, which calculates the power offset under temperature change;

[0052] Electric vehicle charging probability model, which generates charging spatio-temporal distribution based on user travel big data.

[0053] The planning decision platform includes:

[0054] The capacity optimization module configures the energy storage location and capacity with the goal of reducing the net load peak by 20%;

[0055] The network frame reconstruction algorithm identifies weak links based on the power flow entropy index and generates line capacity increase / segmentation schemes;

[0056] The economic evaluator compares the life cycle cost of the planning scheme with the new energy consumption benefit.

[0057] The capacity optimization module performs:

[0058] When the predicted feeder net load is continuously greater than 90% of the limit, start the energy storage charging instruction;

[0059] When the new energy output drops sharply, call the energy storage discharge and trigger the demand response compensation mechanism.

[0060] The dynamic feedback module includes:

[0061] The error tracking unit calculates the deviation rate of the actual net load and the predicted value;

[0062] The model self-learning unit automatically adjusts the load decomposition weight coefficient when the deviation rate is greater than 10%;

[0063] The planning verification interface feeds back the running voltage out-of-limit and equipment overload events to the planning decision platform.

[0064] The system configures a multi-scenario simulator to generate flexible planning schemes under extreme high temperature, new energy off-grid and other scenarios, and output risk level assessment reports.

[0065] The basic load prediction unit accesses high-frequency data from smart meters and identifies impact loads such as industrial electric arc furnaces and rolling mills through a load characteristic waveform library.

[0066] The proposed grid reconfiguration algorithm introduces a dynamic capacity ratio index to constrain the transformer load rate within a safe range of 40%-80%, and prioritizes the deployment of distributed power supply access points.

[0067] The system communicates with the power grid dispatch CMS system and the marketing management system to achieve closed-loop control of "prediction-planning-operation-verification".

Claims

1. A hierarchical refined net load forecasting and lean distribution network planning system, characterized in that, include: The load forecasting engine generates net load curves for distribution areas, feeders, and substations according to spatial hierarchy. The planning and decision-making platform outputs optimized grid structure solutions based on forecast results. A dynamic feedback module collects operational data in real time to correct the prediction and planning model; wherein, the load prediction engine includes: Basic load forecasting unit, decomposes rigid electrical loads; A flexible load response model quantifies the impact of electricity price / temperature on adjustable loads; A new energy output corrector that integrates meteorological satellite data to calibrate photovoltaic / wind power forecasts.

2. The hierarchical refined net load forecasting and lean distribution network planning system according to claim 1, characterized in that: The load forecasting engine performs hierarchical forecasting: Step S1: Non-intrusive identification technology is used at the transformer substation level to separate flexible loads such as air conditioners and charging piles; Step S2: Overlay the net load of the transformer area at the feeder level and calculate the line loss by associating the topology impedance; Step S3: Aggregate feeder data at the substation level and introduce transfer load constraints to generate regional net load curves.

3. The hierarchical refined net load forecasting and lean distribution network planning system according to claim 2, characterized in that: The flexible load response model includes: Equivalent thermal parameter model of air conditioning cluster, calculate power offset under temperature change; Electric vehicle charging probability model, which generates the spatiotemporal distribution of charging based on user travel big data.

4. The hierarchical refined net load forecasting and lean distribution network planning system according to claim 1, characterized in that: The planning decision-making platform includes: The capacity optimization module configures energy storage locations and capacities with the goal of reducing peak net load by 20%. A grid reconstruction algorithm identifies weak links based on the power flow entropy index and generates line capacity expansion / segmentation schemes. An economic evaluator compares the total life-cycle cost of a planning scheme with the benefits of renewable energy consumption.

5. The hierarchical refined net load forecasting and lean distribution network planning system according to claim 4, characterized in that: The capacity optimization module performs the following: When it is predicted that the net load of the feeder will continue to be greater than 90% of the limit, the energy storage charging command will be initiated. When the output of new energy sources suddenly drops, the energy storage is activated to discharge and trigger the demand response compensation mechanism.

6. The hierarchical refined net load forecasting and lean distribution network planning system according to claim 1, characterized in that: The dynamic feedback module includes: The error tracking unit calculates the deviation rate between the actual net load and the predicted value. The model has a self-learning unit that automatically adjusts the load decomposition weight coefficients when the deviation rate is greater than 10%. The planning verification interface feeds back events such as operating voltage exceeding limits and equipment overload to the planning decision platform.

7. The hierarchical refined net load forecasting and lean distribution network planning system according to claim 1, characterized in that: The system is configured with multi-scenario simulators to generate flexible planning schemes for scenarios such as extreme high temperatures and grid disconnection of new energy sources, and outputs risk level assessment reports.

8. The hierarchical refined net load forecasting and lean distribution network planning system according to claim 1, characterized in that: The basic load prediction unit accesses high-frequency data from smart meters and identifies impact loads such as industrial electric arc furnaces and rolling mills through a load characteristic waveform library.

9. The hierarchical refined net load forecasting and lean distribution network planning system according to claim 4, characterized in that: The proposed grid reconfiguration algorithm introduces a dynamic capacity ratio index to constrain the transformer load rate within a safe range of 40%-80%, and prioritizes the deployment of distributed power supply access points.

10. The hierarchical refined net load forecasting and lean distribution network planning system according to claim 1, characterized in that: The system communicates with the power grid dispatch CMS system and the marketing management system to achieve closed-loop control of "prediction-planning-operation-verification".