CN115907228A - Short-term power load prediction analysis method based on PSO-LSSVM - Google Patents
Short-term power load prediction analysis method based on PSO-LSSVM Download PDFInfo
- Publication number
- CN115907228A CN115907228A CN202211737565.9A CN202211737565A CN115907228A CN 115907228 A CN115907228 A CN 115907228A CN 202211737565 A CN202211737565 A CN 202211737565A CN 115907228 A CN115907228 A CN 115907228A
- Authority
- CN
- China
- Prior art keywords
- data
- lssvm
- load
- prediction
- pso
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Abstract
The invention discloses a short-term power load prediction analysis method based on a PSO-LSSVM. The method comprises the following steps: acquiring original data and judging the data; step two: preprocessing original data; step three: based on the LSSVM technology, selecting a load prediction decision according to the optimized effective data and the load prediction requirement; step four: predicting the load at a specific time based on the steps; step five: verifying the data, and finally outputting prediction result data of the power load; step six: the weight of each index is determined by the information quantity provided by each index by using an entropy weight method. The method has the advantages of simple operation, high intelligent degree, short time consumption and high prediction precision.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to a short-term power load prediction analysis method based on a PSO-LSSVM.
Background
The power load data is one of core data of a power network and is related to professional fields of metering, marketing, scheduling and the like of a power grid. Due to problems such as metering devices, time-sharing power data may be missing or abnormal. On the other hand, a user has a relatively fixed power consumption behavior pattern, and an unexpected event may generate a power consumption behavior different from a normal state, and the power load data at this time may be regarded as abnormal data, which may interfere with applications such as user behavior and user profile analysis. Therefore, an effective method is required to be found for identifying the abnormal data of the power load.
The load prediction is to determine load data of a certain future moment under the condition of meeting a certain precision requirement according to a plurality of factors such as the operation characteristic, capacity increase decision, natural conditions, social influence and the like of a system, wherein the load refers to power demand (power) or power consumption, the load prediction is an important content in the economic dispatching of a power system and is an important module of an Energy Management System (EMS), the power load prediction is one of important work of a power department, the accurate load prediction can economically and reasonably arrange the start and stop of a generator set in a power grid, the safety and stability of the operation of the power grid are kept, unnecessary rotation reserve capacity is reduced, a unit maintenance plan is reasonably arranged, the normal production and life of the society are guaranteed, the power generation cost is effectively reduced, and the economic benefit and the social benefit are improved;
currently, most of the power load forecasting methods are manual forecasting methods, the method needs to manually collect power load data before a day to be forecasted, forecasting is carried out according to the power load data, and the power load data of the day to be forecasted are obtained, the method is complex in operation, low in intellectualization, long in time consumption, influenced in forecasting precision and low in popularization value; therefore, it is necessary to develop a short-term power load prediction analysis method based on the PSO-LSSVM, which is simple in operation, high in intelligence degree, short in time consumption, and high in prediction accuracy.
Disclosure of Invention
The invention aims to provide a short-term power load prediction analysis method based on a PSO-LSSVM, which is simple to operate, high in intelligent degree, short in time consumption and high in prediction accuracy, introduces a self-adaptive step length on the basis of keeping the advantages of the PSO, enables the PSO to find out more optimal particles on a speed gradient, optimizes the parameters of the LSSVM regression estimation method by the self-adaptive PSO to obtain the self-adaptive PSO-LSSVM, and finally applies the method to short-term power load prediction.
In order to realize the purpose, the technical scheme of the invention is as follows: a short-term power load prediction analysis method based on PSO-LSSVM is characterized in that: comprises the following steps of (a) preparing a solution,
the method comprises the following steps: acquiring original data and judging the data;
step two: preprocessing original data;
step three: based on the LSSVM technology, selecting a load prediction decision according to the optimized effective data and the load prediction requirement;
step four: predicting the load at a specific time based on the steps;
step five: and verifying the data, and finally outputting the prediction result data of the power load.
In the above technical solution, in the first step, the data is discriminated by the specific method:
the method adopts a robust estimation method, the theoretical basis of which is that a certain pollution distribution of data distribution load is assumed, bad data is removed by using a learning vector quantization method, the data is regarded as a plurality of component quantities, one component in a certain vector is bad data, the whole vector is removed, so that the bad data is accurately positioned, and the suspicious point is judged by comparing the load change rates of two points on a date to be detected within a normal unit through counting the range of the load change rate between two times every day historically.
In the above technical solution, in the second step, the raw data includes the operating characteristics and the natural environment of the power system, and specifically includes weather factors, date types and special events, where the weather factors include temperature, humidity and sunshine intensity, the date types include working days and holidays, and the special events include system faults, natural disasters and political events.
In the above technical solution, in the operation process, since the unit and the magnitude of each index are different, a certain characteristic index is abnormal and prominent, and in order to eliminate the difference, normalization processing, including normalization of load data, division and normalization of date type, and normalization of weather conditions, is performed on each characteristic index.
In the above technical solution, in the second step, the pre-processing of the original data includes supplementing and correcting the historical load value before use, selecting the data of the same type near the date, using the following formula,
x(d,t)＝z 1 y(d-1，t)+z 2 y(d+1,t)
wherein x (d, t) represents the load value of a time sampling point t on the d th day, z1 and z2 are weights of weighted average, 0.5 is taken, and when sampling is performed 1 time per hour, x (d-1,t) and x (d +1,t) represent the load sampling values of the corresponding time of the previous day and the next day.
In the above technical solution, in the third step, a load prediction decision is selected, and the specific method is as follows:
step S31: firstly, establishing an LSSVM model, selecting a load prediction decision according to the optimized effective data and the load prediction requirement:
the LSSVM model is described as follows: given a set of samples x i ,y i } m i＝1 X i ∈R n ，y i ∈R n Is the corresponding output value in the sample, and the sample data is mapped to a higher dimensional space by using a nonlinear function phi, expressed by a method of approximating linear approximation,
f(x)＝ω T φ(x)+a
where ω is a weight, a is a threshold;
because the short-term load can be influenced by the factors of temperature, humidity and rainfall, the scheme mainly considers two factors with large influence on the temperature and the humidity to establish a prediction model, takes the data of the first 30 days as a PSO-LSSVM as a training sample, takes the data of the last 30 days as a test sample, includes the daily maximum air temperature, the daily minimum air temperature, the daily average air temperature and the humidity value in the training input sample and the test input sample as characteristic indexes, and outputs the load value as a sample point;
step S32: by using the algorithm of the extreme learning machine,
for a single hidden layer neural network with an N-l-m structure, given N training samples, { (x) i ，t i )} N i＝1 Input data is x i ＝[X i1 ,X i2 ，…X in ] T ∈R n Output value of t i ＝[t i1 ,t i2 …t im ] T ∈R n The output of the KELM network is,
in the formula, y i ∈R n Is the output value of the network, beta is the output weight between the hidden node and the output layer, g i (ω i X j +b i ) Activation function for i-th hidden layer node，ω i To connect the weight between the ith hidden layer node and the input node, b i Is the threshold of the ith occlusion layer of the network.
Step S33: the prediction result is evaluated by using the coverage rate of the prediction interval and the average bandwidth of the prediction interval, the formula is as follows,
in the formula, N t To predict the number of samples, k i (α) Is a Boolean quantity, L t (α) Lower limit of prediction interval, U, at confidence 1-alpha t (α) Is the lower limit of the prediction interval at confidence 1-alpha.
In the above technical solution, in the fourth step, the load at a specific time is predicted, and the specific method is as follows:
step S41: carrying out PSO-LSSVM optimization, evaluating the population and generating a new population;
step S42: whether the detection is finished or not is finished, and the found optimal position of the particles, namely the optimal parameter vector is endowed to the LSSVM model;
step S43: training an LSSVM model by using effective data, and obtaining a recovery function formula by using the solved optimal solution parameters;
step S44: and predicting the power load at a certain future moment by using the prediction training sample and the LSSVM, and finally correcting the predicted value.
In the above technical solution, in the fourth step, in the Particle Swarm Optimization (PSO), in order to obtain the optimal values of γ and σ, the optimal fitness is obtained by calculating the objective function, and it is proposed to introduce an adaptive mutation operator to mutate the particles that have been trapped in the local optimization, so as to improve the optimization performance of the algorithm, that is, regarding the hyper-parameter γ and σ in the kernel function, which need to be selected in the LSSVM, as two particles, and continuously updating the positions and velocities of the two particles until the optimal values are obtained.
In the above technical solution, in the step five, the data is verified, and the specific method is as follows:
determining index weight by using an entropy weight method through the information quantity provided by each index, and finally outputting prediction result data of the power load, wherein the weights of the KELM and the LSSVM in the combined prediction model are determined to be omega 1 and omega 2 through objective evaluation of the entropy weight method, so that a final prediction interval is established to be omega 1
The expression is as follows:
in the formula of U 1 ,L 1 Upper and lower bounds of KLEM separate prediction interval; u shape 2 ，L 2 The upper limit and the lower limit of the LSSVM single prediction interval.
The PSO (Particle Swarm Optimization) in the PSO-LSSVM isParticle swarm optimization algorithm(ii) a The LSSVM is a least squares support vector machine.
The invention has the following beneficial effects:
1. the invention discloses a short-term power load prediction analysis method based on a PSO-LSSVM (particle swarm optimization-least squares support vector machine), which introduces a self-adaptive step length on the basis of keeping the advantages of the PSO, so that the PSO finds out more optimal particles on a velocity gradient, the self-adaptive PSO optimizes the parameters of an LSSVM regression estimation method to obtain the self-adaptive PSO-LSSVM, and finally the method is applied to short-term power load prediction.
2. The short-term power load prediction analysis method based on the PSO-LSSVM has feasibility and effectiveness, the average error of model prediction is 0.85%, the accuracy is higher, and the method is suitable for power load prediction.
Of course, it is not necessary for any product to practice the invention to achieve all of the above-described advantages at the same time.
Drawings
FIG. 1 is a first flowchart of the operation of the short-term power load prediction analysis method based on PSO-LSSVM of the present invention;
FIG. 2 is a second operation flow chart of the short-term power load prediction analysis method based on the PSO-LSSVM of the present invention;
fig. 3 is an architecture diagram of an operation terminal of the short-term power load prediction analysis method based on the PSO-LSSVM of the present invention.
Detailed Description
The embodiments of the present invention will be described in detail with reference to the accompanying drawings, which are not intended to limit the present invention, but are merely exemplary. While the advantages of the invention will be apparent and understood from the description.
With reference to the accompanying drawings: the invention relates to a short-term power load prediction analysis method based on a PSO-LSSVM, which comprises the following steps:
s1: acquiring original data and judging the data; when data are judged, a robust estimation method is adopted, the theoretical basis is that certain pollution distribution of data distribution load is assumed, bad data are removed by a learning vector quantization method, the data are regarded as a plurality of component quantities, one component in a certain vector is bad data, the whole vector is removed, accordingly, the bad data are accurately positioned, and the load change rates of two points on a date to be detected are compared in a normal unit through counting the range of the load change rate between two moments every day historically, so that the suspicious point is judged.
S2: preprocessing raw data, wherein the preprocessing of the raw data comprises the steps of supplementing and correcting historical load values before use, selecting data of the same type near dates, and adopting the following formula, wherein x (d, t) = z 1 y(d-1，t)+z 2 y (d +1,t); wherein x (d, t) represents the load value of a time sampling point t on the d th day, z1 and z2 are weights of weighted average, 0.5 is taken, and when sampling is carried out for 1 time per hour, x (d-1,t) and x (d +1,t) represent load sampling values of corresponding moments on the previous day and the next day; the original data comprise the operating characteristics and the natural environment of the power system, and specifically comprise weather factors, date types and special events, wherein the weather factors comprise temperature, humidity and sunshine intensity, the date types comprise working days and holidays, and the special events comprise system faults, natural disasters and political events; in the operation process, as the unit and the magnitude of each index are different, a certain characteristic index is abnormal and prominent in order to eliminateIn addition to the difference, each characteristic index is subjected to normalization processing, including normalization of load data, division and normalization of date types, and weather conditions.
S3: based on the LSSVM technology, selecting a load prediction decision according to the optimized effective data and the load prediction demand, wherein the method comprises the following steps of firstly establishing an LSSVM model, and the LSSVM model is described as follows: given a set of samples x i ,y i } m i＝1 X i ∈R n ，y i e.Rn is the corresponding output value in the sample, and the sample data is mapped to a higher dimensional space by utilizing a nonlinear function phi and expressed by an approximate linear approximation method,
f(x)＝ω T φ(x)+a
where ω is a weight and a is a threshold; because the short-term load can be influenced by the factors of temperature, humidity and rainfall, the scheme mainly considers two factors with large influence of temperature and humidity to establish a prediction model, takes the data of the first 30 days as a PSO-LSSVM as a training sample, takes the data of the last 30 days as a test sample, includes the daily maximum air temperature, the daily minimum air temperature, the daily average air temperature and humidity value and the daily load data as characteristic indexes in the training input sample and the test input sample, and outputs the load value as a sample point;
the scheme utilizes the coverage rate of the prediction interval and the average bandwidth of the prediction interval to evaluate the prediction result, and has the following formula,
in the formula, N t To predict the number of samples, k i (α) Is a Boolean quantity, L t (α) Lower limit of prediction interval, U, at confidence 1-alpha t (α) Is the lower limit of the prediction interval at confidence 1-alpha.
S4: based on the steps, the load at a specific moment is predicted, including (1) PSO-LSSVM optimization is carried out, the population is evaluated, and a new population is generated; (2) whether the detection is finished or not is judged, and the optimal position of the searched particles, namely the optimal parameter vector, is given to the LSSVM model is finished; (3) training an LSSVM model by using the effective data, and obtaining a recovery function formula by using the solved optimal solution parameter; (4) predicting the power load at a certain moment in the future by using the prediction training sample and the LSSVM, and finally correcting the predicted value; in a Particle Swarm Optimization (PSO), in order to obtain the optimal values of gamma and sigma, the optimal fitness is obtained by calculating an objective function, a self-adaptive mutation operator is introduced to mutate the particles which are trapped into local optimization, the optimization performance of the PSO is improved, namely, the hyper-parameter gamma and the sigma in a kernel function which need to be selected in an LSSVM are regarded as two particles, and the positions and the speeds of the two particles are continuously updated until the optimal values are obtained.
S5: verifying the data, and finally outputting prediction result data of the power load;
the weights of the KELM and the LSSVM in the combined prediction model are determined to be omega 1 and omega 2 respectively through objective evaluation of an entropy weight method, so that a final prediction interval is established to be omega 2
Is expressed as
In the formula of U 1 ,L 1 Is the upper and lower limit of the KLEM separate prediction interval; u shape 2 ，L 2 The upper limit and the lower limit of the LSSVM single prediction interval.
The PSO algorithm flow in the scheme is as follows:
step1: initializing the speed and position of population particles, if the search space is d-dimensional, indicating that each particle has d variables, initializing the current historical optimal position p best Taking the global optimal position g of the particle swarm as the initial position best The optimum value of (1);
step2: estimating the fitness of each particle, recording the optimal position and the fitness of each particle and taking the optimal position and the fitness as the position of the current population; step3: step2 is executed to continuously change and update the position and the speed of the particles; step4:current p best The corresponding fitness is compared with the updated and adjusted particle fitness, and if the fitness is better than the current fitness, the updated particle position is taken as P best ；
Step5: current g best The corresponding fitness is compared with the fitness of each particle, and if the fitness is better than the current fitness, the value of the gbest is updated; step6: and (5) judging whether a termination condition is met, if so, ending, and if not, continuing to loop Step3 until the condition is met.
In the scheme of the invention, the influence factors of the short-term power load comprise (1) load influence factors: and inputting the influence factors of temperature, humidity, rainfall, wind speed, date type and the like of each point with strong correlation, and then matching the date with the meteorological characteristic date. If the data base is lack of meteorology, a linear interpolation method is adopted to make up the meteorology characteristics of each point; (2) daily electricity consumption prediction: inputting data of characteristic parameters such as temperature, humidity, rainfall at each point of the day, the highest temperature, the lowest temperature, the average temperature, the date type and the like at the day; (3) monthly electricity sales (consumption) prediction: inputting data such as electricity price, temperature of each section, holiday days and the like; (5) and predicting annual electricity sale (use) quantity: inputting various influence factors such as population, GDP, consumption coefficient, price index, power consumption in the whole society, per-capita output value, per-capita power consumption and unit consumption of the output value, introducing appropriate influence factors through Glange causal test due to excessive parameters, and introducing principal component analysis and extraction of main influence factors;
as shown in fig. 3, in this solution, the short-term power load prediction analysis terminal based on the PSO-LSSVM includes an application layer, a platform layer and a resource layer, the application function of the application layer includes a load prediction function and a query function, the platform layer includes an interface, a data processing port, a data storage port and a tool, and is used for receiving and storing data and belonging to the data, and the resource layer includes historical load data and historical weather data, and is used for acquiring data information.
In the description herein, references to the description of "one embodiment," "an example," "a specific example" or the like are intended to mean that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above do not necessarily refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
The preferred embodiments of the invention disclosed above are intended to be illustrative only. The preferred embodiments are not intended to be exhaustive or to limit the invention to the precise embodiments disclosed. Obviously, many modifications and variations are possible in light of the above teaching. The embodiments were chosen and described in order to best explain the principles of the invention and the practical application, to thereby enable others skilled in the art to best understand the invention for and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.
Other parts not described belong to the prior art.
Claims (8)
1. A short-term power load prediction analysis method based on PSO-LSSVM is characterized in that: comprises the following steps of (a) carrying out,
the method comprises the following steps: acquiring original data and judging the data;
step two: preprocessing original data;
step three: based on the LSSVM technology, selecting a load prediction decision according to the optimized effective data and the load prediction requirement;
step four: predicting the load at a specific time based on the steps;
step five: and verifying the data, and finally outputting the prediction result data of the power load.
2. The PSO-LSSVM-based short-term power load prediction analysis method of claim 1, wherein: in step one, the raw data comprises the operating characteristics and the natural environment of the power system, specifically comprising weather factors, date types and special events, wherein the weather factors comprise temperature, humidity and sunshine intensity, the date types comprise working days and holidays, and the special events comprise system faults, natural disasters and political events.
3. The PSO-LSSVM-based short-term power load prediction analysis method according to claim 1 or 2, wherein: in the first step, data discrimination is performed, and the specific method is as follows:
the method adopts a robust estimation method, the theoretical basis of which is to assume that the data distribution loads a certain pollution distribution, and eliminate bad data by using a learning vector quantization method, the data is regarded as a plurality of components, one component in a certain vector is bad data, the whole vector is eliminated, so that the bad data is accurately positioned, and the suspicious point is judged by comparing the load change rates of the two points on the date to be detected within a normal unit through counting the range of the load change rate between two moments every day historically.
4. The PSO-LSSVM based short-term power load prediction analysis method of claim 3, wherein: in the second step, the pre-processing of the original data comprises the steps of supplementing and correcting the historical load value before use, selecting the data with the same type near the date, adopting the formula as follows,
x(d,t)＝z 1 y(d-1，t)+z 2 y(d+1,t)
wherein x (d, t) represents the load value of a time sampling point at the time t on the day d, z1 and z2 are weights of weighted average, 0.5 is taken, and when sampling is carried out for 1 time per hour, x (d-1,t) and x (d +1,t) represent load sampling values at corresponding moments on the previous day and the next day;
in the operation process, as the unit and the magnitude of each index are different, a certain characteristic index is abnormal and prominent, and in order to eliminate the difference, normalization processing including load data normalization, date type division and normalization and weather condition normalization is carried out on each characteristic index.
5. The PSO-LSSVM based short-term power load prediction analysis method of claim 4, wherein: in the third step, a load prediction decision is selected, and the specific method comprises the following steps:
step S31: firstly, establishing an LSSVM model:
the LSSVM model is described as follows: given a set of samples x i ,y i } m i＝1 X i ∈R n ，y i ∈R n Is the corresponding output value in the sample, and the sample data is mapped to a higher dimensional space by using a nonlinear function phi, expressed by a method of approximating linear approximation,
f(x)＝ω T φ(x)+a
where ω is a weight and a is a threshold;
because the short-term load is influenced by the factors of temperature, humidity and rainfall, a prediction model is established by considering two factors with large influences of temperature and humidity, the data of the first 30 days are taken as PSO-LSSVM as training samples, the data of the last 30 days are taken as test samples, the training input samples and the test input samples comprise daily maximum air temperature, daily minimum air temperature, daily average air temperature and humidity values and the daily load data as characteristic indexes, and the load values are output as sample points;
step S32: by using the algorithm of the extreme learning machine,
for a single hidden layer neural network with an N-l-m structure, given N training samples,
input data is x i ＝[X i1, X i2，… X in ] T ∈R n Output value of t i ＝[t i1 ,t i2… t im ] T ∈R n The output of the KELM network is,
in the formula, y i ∈R n Is the output value of the network, beta is the output weight between the hidden node and the output layer, g i (ω i X j +b i ) Is composed ofActivation function of i-th hidden layer node, ω i To connect the weight between the ith hidden layer node and the input node, b i A threshold value of the ith shadow layer of the network;
step S33: the prediction result is evaluated by using the coverage rate of the prediction interval and the average bandwidth of the prediction interval, the formula is as follows,
in the formula, N t To predict the number of samples, k i (α) Is a Boolean quantity, L t (α) Lower limit of prediction interval, U, at confidence 1-alpha t (α) Is the lower limit of the prediction interval at confidence 1- α.
6. The PSO-LSSVM based short-term power load prediction analysis method of claim 5, wherein: in the fourth step, the load at a specific time is predicted, and the specific method is as follows:
step S41: carrying out PSO-LSSVM optimization, evaluating the population and generating a new population;
step S42: whether the detection is finished or not is finished, and the found optimal position of the particles, namely the optimal parameter vector is endowed to the LSSVM model;
step S43: training an LSSVM model by using effective data, and obtaining a recovery function formula according to the solved optimal solution parameter;
step S44: and predicting the power load at a certain moment in the future by using the prediction training sample and the LSSVM, and finally correcting the predicted value.
7. The PSO-LSSVM based short-term power load prediction analysis method of claim 6, wherein: in the fourth step, in the particle swarm optimization, in order to obtain the optimal values of gamma and sigma, the optimal fitness is obtained by calculating a target function, a self-adaptive mutation operator is introduced to mutate the particles which are trapped into the local optimization, the optimization performance of the particle swarm optimization is improved, namely, the hyper-parameter gamma and the sigma in the kernel function which need to be selected in the LSSVM are regarded as two particles, and the positions and the speeds of the two particles are continuously updated until the optimal values of the two particles are obtained.
8. The PSO-LSSVM-based short-term power load prediction analysis method of claim 7, wherein: in the fifth step, the data is verified, and the specific method comprises the following steps:
determining index weight through the information quantity provided by each index by using an entropy weight method, and finally outputting prediction result data of the power load, wherein weights of the KELM and the LSSVM in the combined prediction model are determined to be omega 1 and omega 2 respectively through objective evaluation of the entropy weight method, so that a final prediction interval is established to be omega 1
The expression is as follows:
in the formula of U 1 、L 1 Upper and lower limits of KLEM separate prediction intervals; u shape 2 、L 2 The upper limit and the lower limit of the LSSVM single prediction interval.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211737565.9A CN115907228A (en) | 2022-12-30 | 2022-12-30 | Short-term power load prediction analysis method based on PSO-LSSVM |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211737565.9A CN115907228A (en) | 2022-12-30 | 2022-12-30 | Short-term power load prediction analysis method based on PSO-LSSVM |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115907228A true CN115907228A (en) | 2023-04-04 |
Family
ID=86483990
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211737565.9A Pending CN115907228A (en) | 2022-12-30 | 2022-12-30 | Short-term power load prediction analysis method based on PSO-LSSVM |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115907228A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117335409A (en) * | 2023-10-26 | 2024-01-02 | 河北建投电力科技服务有限公司 | Power consumer load prediction system based on artificial intelligence |
| CN117335409B (en) * | 2023-10-26 | 2024-04-19 | 河北建投电力科技服务有限公司 | Power consumer load prediction system based on artificial intelligence |
-
2022
- 2022-12-30 CN CN202211737565.9A patent/CN115907228A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117335409A (en) * | 2023-10-26 | 2024-01-02 | 河北建投电力科技服务有限公司 | Power consumer load prediction system based on artificial intelligence |
| CN117335409B (en) * | 2023-10-26 | 2024-04-19 | 河北建投电力科技服务有限公司 | Power consumer load prediction system based on artificial intelligence |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110097297B (en) | Multi-dimensional electricity stealing situation intelligent sensing method, system, equipment and medium | |
| Dong et al. | Wind power day-ahead prediction with cluster analysis of NWP | |
| CN112508275B (en) | Power distribution network line load prediction method and equipment based on clustering and trend indexes | |
| CN109919353B (en) | Distributed photovoltaic prediction method of ARIMA model based on spatial correlation | |
| CN108388962B (en) | Wind power prediction system and method | |
| CN111369070B (en) | Multimode fusion photovoltaic power prediction method based on envelope clustering | |
| CN105373857A (en) | Photovoltaic power station irradiance prediction method | |
| CN111753893A (en) | Wind turbine generator power cluster prediction method based on clustering and deep learning | |
| CN113554466B (en) | Short-term electricity consumption prediction model construction method, prediction method and device | |
| CN111178585A (en) | Fault reporting amount prediction method based on multi-algorithm model fusion | |
| CN115358347B (en) | Method for predicting remaining life of intelligent electric meter under different subsystems | |
| CN109583645A (en) | A kind of public building short-term load forecasting method | |
| CN112990587A (en) | Method, system, equipment and medium for accurately predicting power consumption of transformer area | |
| CN110210670A (en) | A kind of prediction technique based on power-system short-term load | |
| CN115545333A (en) | Method for predicting load curve of multi-load daily-type power distribution network | |
| CN112288157A (en) | Wind power plant power prediction method based on fuzzy clustering and deep reinforcement learning | |
| CN111882114A (en) | Short-term traffic flow prediction model construction method and prediction method | |
| CN115600729A (en) | Grid load prediction method considering multiple attributes | |
| CN112036598A (en) | Charging pile use information prediction method based on multi-information coupling | |
| CN115099511A (en) | Photovoltaic power probability estimation method and system based on optimized copula | |
| CN108830405B (en) | Real-time power load prediction system and method based on multi-index dynamic matching | |
| CN109978236A (en) | A kind of small power station's short term power fining prediction technique based on feature combination | |
| Ibrahim et al. | LSTM neural network model for ultra-short-term distribution zone substation peak demand prediction | |
| CN116960962A (en) | Mid-long term area load prediction method for cross-area data fusion | |
| CN116307076A (en) | Industrial park energy efficiency management and control method based on Internet of things |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |