CN-121984079-A - Thermal power generating unit load prediction method, thermal power generating unit load prediction device, computer equipment, readable storage medium and program product

CN121984079ACN 121984079 ACN121984079 ACN 121984079ACN-121984079-A

Abstract

The application relates to a thermal power unit load prediction method, a thermal power unit load prediction device, computer equipment, a readable storage medium and a program product, and relates to the technical field of unit load prediction. The method comprises the steps of responding to a load prediction instruction for a thermal power unit, obtaining current operation state characteristics of the thermal power unit, detecting similarity between the operation state characteristics and working condition characteristics of candidate operation working conditions, performing model parameter adjustment matching according to the similarity to obtain model parameter adjustment quantity matched with the similarity, obtaining historical load sample data of the thermal power unit under the current shift, performing parameter adjustment on a pre-trained load prediction model based on the historical load sample data and the model parameter adjustment quantity to obtain a target prediction model matched with the current operation state of the thermal power unit, and analyzing real-time operation data of the thermal power unit through the target prediction model to obtain a load prediction result of the thermal power unit. By adopting the method, the accuracy of the power grid dispatching decision can be ensured.

Inventors

CHEN HAOYU
HAO XIAOLIANG
LIU ZHENG
CHU JINGCHUN
XIAO BAOLING
WANG LIJIE
SUN HONGYANG

Assignees

国家能源集团新能源技术研究院有限公司
国家能源投资集团有限责任公司
国能长源能源销售有限公司

Dates

Publication Date: 20260505
Application Date: 20251208

Claims (10)

1. A thermal power generating unit load prediction method, the method comprising: Responding to a load prediction instruction aiming at a thermal power unit, and acquiring the current running state characteristics of the thermal power unit; Detecting the similarity between the operating state characteristics and the operating condition characteristics of the candidate operating conditions; According to the similarity, performing model parameter adjustment matching to obtain a model parameter adjustment quantity matched with the similarity; Acquiring historical load sample data of the thermal power generating unit under the current shift; Based on the historical load sample data and the model parameter adjustment quantity, performing parameter adjustment on a pre-trained load prediction model to obtain a target prediction model which is adaptive to the current running state of the thermal power generating unit; and analyzing the real-time operation data of the thermal power unit through the target prediction model to obtain a load prediction result of the thermal power unit.
2. The method of claim 1, wherein the detecting a similarity between the operating state characteristic and an operating state characteristic of a candidate operating state comprises: detecting a difference value between the operating state characteristic and the operating condition characteristic of the candidate operating condition; Carrying out weighted summation on the difference values to obtain weighted distances between the running state characteristics and the working condition characteristics of the candidate running working conditions; And determining the similarity between the operating state characteristics and the operating condition characteristics of the candidate operating conditions based on the weighted distance.
3. The method of claim 1, wherein said performing model parameter adjustment matching according to the similarity to obtain a model parameter adjustment matching with the similarity comprises: if the similarity is not smaller than a preset similarity threshold, inquiring a model parameter adjustment quantity matched with the candidate operation working condition; and if the similarity is smaller than the preset similarity threshold, generating a corresponding model parameter adjustment quantity according to the running state characteristics.
4. The method of claim 1, wherein the pre-trained load prediction model comprises a lightweight adaptation layer, wherein the parameter adjusting the pre-trained load prediction model based on the historical load sample data and the model parameter adjustment amount comprises: based on the model parameter adjustment amount, adjusting the model parameters of the lightweight adaptation layer to obtain a lightweight adaptation layer after parameter adjustment; and taking the historical load sample data as a training sample, and carrying out gradient update on the light-weight adaptive layer after parameter adjustment.
5. The method of claim 1, wherein the target prediction model comprises a time domain channel and a frequency domain channel, wherein the analyzing the real-time operation data of the thermal power unit by the target prediction model to obtain the load prediction result of the thermal power unit comprises: Analyzing real-time operation data of the thermal power unit through the time domain channel to obtain a first prediction result reflecting the long-term load change trend of the thermal power unit; analyzing the real-time operation data through the frequency domain channel to obtain a second prediction result reflecting the short-term load disturbance of the thermal power generating unit; and fusing the first prediction result and the second prediction result to obtain a load prediction result of the thermal power generating unit.
6. The method according to claim 1, wherein the method further comprises: acquiring a plurality of operation state feature vectors of the thermal power generating unit in a historical operation period; Detecting weighted distances among the operation state feature vectors, and grouping the operation state feature vectors based on the weighted distances to obtain candidate operation conditions; Detecting the business risk level of the candidate operation working condition, and constructing the candidate operation working condition into a training task based on the business risk level; And training the initial prediction model according to the training task to obtain a load prediction model.
7. A thermal power generating unit load prediction apparatus, the apparatus comprising: The characteristic acquisition module is used for responding to a load prediction instruction for the thermal power unit and acquiring the current running state characteristic of the thermal power unit; the similarity detection module is used for detecting similarity between the operating state characteristics and the operating condition characteristics of the candidate operating conditions; The parameter adjustment matching module is used for carrying out model parameter adjustment matching according to the similarity to obtain a model parameter adjustment quantity matched with the similarity; The load data acquisition module is used for acquiring historical load sample data of the thermal power generating unit under the current shift; the model updating module is used for carrying out parameter adjustment on the pre-trained load prediction model based on the historical load sample data and the model parameter adjustment quantity to obtain a target prediction model which is matched with the current running state of the thermal power generating unit; And the load prediction module is used for analyzing the real-time operation data of the thermal power unit through the target prediction model to obtain a load prediction result of the thermal power unit.
8. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the steps of the method of any of claims 1 to 6 when the computer program is executed.
9. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any of claims 1 to 6.
10. A computer program product comprising a computer program, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any of claims 1 to 6.

Description

Thermal power generating unit load prediction method, thermal power generating unit load prediction device, computer equipment, readable storage medium and program product Technical Field The application relates to the technical field of unit load prediction, in particular to a thermal power unit load prediction method, a thermal power unit load prediction device, computer equipment, a readable storage medium and a program product. Background At present, a technical system of 'unit independent modeling' and 'periodic retraining' is generally adopted for load prediction of thermal power units, namely, each thermal power unit is independently modeled, and periodic retraining and updating are carried out. However, in this way, the model parameters need to rely on long-period samples to converge, which means that once the thermal power unit overhauls or the fire coal condition is suddenly changed, the complete training process must be re-entered, the model empty window period can last for a plurality of hours or even a plurality of days, during which the load prediction function can fail, and the power grid scheduling decision is easy to deviate. Disclosure of Invention In view of the foregoing, it is desirable to provide a thermal power generating unit load prediction method, apparatus, computer device, computer readable storage medium, and computer program product that can ensure accuracy of power grid dispatching decisions. In a first aspect, the application provides a load prediction method of a thermal power generating unit, comprising the following steps: Responding to a load prediction instruction aiming at the thermal power generating unit, and acquiring the current running state characteristics of the thermal power generating unit; detecting similarity between the operating state characteristics and the operating condition characteristics of the candidate operating conditions; According to the similarity, performing model parameter adjustment matching to obtain a model parameter adjustment quantity matched with the similarity; acquiring historical load sample data of the thermal power generating unit under the current shift; based on historical load sample data and model parameter adjustment quantity, performing parameter adjustment on the pre-trained load prediction model to obtain a target prediction model which is matched with the current running state of the thermal power generating unit; And analyzing real-time operation data of the thermal power unit through the target prediction model to obtain a load prediction result of the thermal power unit. In one embodiment, detecting similarity between an operating state characteristic and an operating state characteristic of a candidate operating state includes: detecting a difference value between the operating state characteristic and the operating condition characteristic of the candidate operating condition; carrying out weighted summation on the difference values to obtain weighted distances between the operating state characteristics and the operating condition characteristics of the candidate operating conditions; and determining the similarity between the operating state characteristics and the operating condition characteristics of the candidate operating conditions based on the weighted distance. In one embodiment, according to the similarity, performing model parameter adjustment matching to obtain a model parameter adjustment amount matched with the similarity, including: If the similarity is not smaller than a preset similarity threshold, inquiring a model parameter adjustment quantity matched with the candidate operation working condition; And if the similarity is smaller than a preset similarity threshold, generating a corresponding model parameter adjustment quantity according to the running state characteristics. In one embodiment, the pre-trained load prediction model comprises a lightweight adaptation layer, and parameter adjustment is performed on the pre-trained load prediction model based on historical load sample data and model parameter adjustment amounts, comprising: based on the model parameter adjustment amount, adjusting the model parameters of the lightweight adaptation layer to obtain a lightweight adaptation layer after parameter adjustment; And taking the historical load sample data as a training sample, and carrying out gradient update on the light-weight adaptation layer after the modulation. In one embodiment, the target prediction model comprises a time domain channel and a frequency domain channel, and the method comprises the steps of analyzing real-time operation data of the thermal power unit through the target prediction model to obtain a load prediction result of the thermal power unit, wherein the load prediction result comprises the following steps: Analyzing real-time operation data of the thermal power unit through the time domain channel to obtain a first prediction result reflecting the long-term load chang