mirror of
https://github.com/WallabyLester/RBF-aPID-Controller.git
synced 2026-05-10 13:09:12 +08:00
Andru Liu
98 lines
3.3 KiB
C++
98 lines
3.3 KiB
C++
#ifndef RBF_MODEL_H
|
|
#define RBF_MODEL_H
|
|
|
|
#include <cmath>
|
|
#include <cstdlib>
|
|
|
|
/**
|
|
* @class RBFModel
|
|
* @brief Radial Basis Function (RBF) Model for function approximation.
|
|
*
|
|
* This class implements an RBF model that evaluates outputs based
|
|
* on a set of radial basis functions centered at specified locations.
|
|
*/
|
|
class RBFModel {
|
|
public:
|
|
|
|
/**
|
|
* @brief Constructor to initialize the RBF model.
|
|
*
|
|
* Initializes the RBF model with a specified number
|
|
* of centers, input dimensions, and the spread (sigma) of the RBFs.
|
|
* Random initialization of the centers can be turned off.
|
|
*
|
|
* @param n_centers The number of radial basis function centers.
|
|
* @param input_dim The dimensionality of the input data.
|
|
* @param sigma The spread of the RBFs (default is 1.0).
|
|
* @param random_centers Boolean to initialize centers randomly (default is true).
|
|
*/
|
|
RBFModel(int n_centers, int input_dim, double sigma = 1.0, bool random_centers = true);
|
|
|
|
/**
|
|
* @brief Destructor to free allocated memory.
|
|
*/
|
|
~RBFModel();
|
|
|
|
/**
|
|
* @brief Predict the RBF output for a given input.
|
|
*
|
|
* @param input A pointer to an array of input values.
|
|
* @return The computed output of the RBF model.
|
|
*/
|
|
double predict(const double* input);
|
|
|
|
/**
|
|
* @brief Adapt weights based on the error and learning rate.
|
|
*
|
|
* @param error The difference between the desired output and the actual output.
|
|
* @param learning_rate The rate at which the weights are adjusted.
|
|
* @param input A pointer to an array of input values used for adaptation.
|
|
*/
|
|
void adapt(double error, double learning_rate, const double* input);
|
|
|
|
/**
|
|
* @brief Train the RBF model using recorded data.
|
|
*
|
|
* @param inputs A pointer to an array of input samples (n_samples x input_dim).
|
|
* @param targets A pointer to an array of target outputs (n_samples).
|
|
* @param n_samples The number of samples to train on.
|
|
* @param epochs The number of training epochs to perform.
|
|
* @param learning_rate The learning rate for weight adaptation.
|
|
*/
|
|
void train(const double* inputs, const double* targets, int n_samples, int epochs, double learning_rate);
|
|
|
|
/**
|
|
* @brief Get the weight at a specific index.
|
|
*
|
|
* @param index The index of the weight to retrieve.
|
|
* @return The weight at the specified index.
|
|
*/
|
|
double get_weight(int index) const;
|
|
|
|
/**
|
|
* @brief Set the weight at a specific index.
|
|
*
|
|
* @param index The index of the weight to set.
|
|
* @param value The new value for the weight.
|
|
*/
|
|
void set_weight(int index, double value);
|
|
|
|
private:
|
|
double** centers; // 2D array for centers
|
|
double* weights; // Array of weights
|
|
int n_centers; // Number of RBF centers
|
|
int input_dim; // Dimension of the input
|
|
double sigma; // Spread of the RBF
|
|
|
|
/**
|
|
* @brief Gaussian function used in RBF evaluation.
|
|
*
|
|
* @param input A pointer to an array of input values.
|
|
* @param center A pointer to an array representing the center of the Gaussian function.
|
|
* @return The output of the Gaussian function.
|
|
*/
|
|
double gaussian(const double* input, const double* center);
|
|
};
|
|
|
|
#endif // RBF_MODEL_H
|