Flex.System (FLex v0.2.2)

View Source

An interface to create a Fuzzy Logic Control System (FLS).

The Fuzzy controllers are very simple conceptually. They consist of an input stage (fuzzification), a processing stage (inference), and an output stage (defuzzification).

Summary

Types

t()

Fuzzy Logic System state.

Functions

backward_pass(pid, desired_output)

Adjust the premise free parameters of the FIS (only avaliable with ANFIS engine), using the following methods

child_spec(init_arg)

Returns a specification to start this module under a supervisor.

compute(pid, input_vector)

Computes the Fuzzy Logic System output for a given input vector.

forward_pass(pid, desired_output)

Adjust the consequent free parameters of the FIS (only avaliable with ANFIS engine), using the following methods

get_state(pid)

Gets the current system state.

hybrid_offline_learning(pid, inputs, targets, epochs)

Adjust the free parameters of the FIS (only avaliable with ANFIS engine), using the following methods

hybrid_online_learning(pid, desired_output)

Adjust the free parameters of the FIS (only avaliable with ANFIS engine), using the following methods

init(params)

Callback implementation for GenServer.init/1.

set_engine_type(pid, type)

Sets the Inference Engine type.

set_learning_rate(pid, learning_rate)

Sets the Learning rate (etha).

start_link(params, opt \\ [])

Spawns a Fuzzy Logic System.

stop(pid)

Types

t()

@type t() :: %Flex.System.State{
  antecedent: [Flex.Variable.t(), ...],
  consequent: Flex.Variable.t(),
  engine_output: Flex.EngineAdapter.engine_state(),
  engine_type:
    Flex.EngineAdapter.Mamdani
    | Flex.EngineAdapter.TakagiSugeno
    | Flex.EngineAdapter.ANFIS,
  initial_gamma: number(),
  learning_rate: number(),
  rules: [Flex.Rule.t(), ...],
  sets_in_rules: list()
}

Fuzzy Logic System state.

  • :rules - (list) A list of rules that defines the behavior of the Fuzzy logic systems.
  • :consequent - Output variable.
  • :antecedent - a list of the input variables.
  • :engine_type - defines the inference engine behavior (default: Mamdini).
  • :sets_in_rules - list of sets involve in the rules (optional, required by ANFIS).
  • :learning_rate - is the speed at which the system parameters are adjusted (ANFIS only).
  • :initial_gamma - is the speed at which the system parameters are adjusted (LSE, ANFIS only).

Functions

backward_pass(pid, desired_output)

@spec backward_pass(
  atom() | pid() | {atom(), any()} | {:via, atom(), any()},
  number()
) ::
  {:ok, number()} | {:error, :einval}

Adjust the premise free parameters of the FIS (only avaliable with ANFIS engine), using the following methods:

  • Learning method: Steepest gradient Backpropagation.
  • Energy function: 0.5 * (target - output)^2

child_spec(init_arg)

Returns a specification to start this module under a supervisor.

See Supervisor.

compute(pid, input_vector)

@spec compute(atom() | pid() | {atom(), any()} | {:via, atom(), any()}, list()) ::
  any()

Computes the Fuzzy Logic System output for a given input vector.

forward_pass(pid, desired_output)

@spec forward_pass(atom() | pid() | {atom(), any()} | {:via, atom(), any()}, number()) ::
  {:ok, number()} | {:error, :einval}

Adjust the consequent free parameters of the FIS (only avaliable with ANFIS engine), using the following methods:

  • Learning method: Steepest gradient Backpropagation.
  • Energy function: 0.5 * (target - output)^2

get_state(pid)

@spec get_state(atom() | pid() | {atom(), any()} | {:via, atom(), any()}) :: t()

Gets the current system state.

hybrid_offline_learning(pid, inputs, targets, epochs)

@spec hybrid_offline_learning(
  atom() | pid() | {atom(), any()} | {:via, atom(), any()},
  list(),
  list(),
  number()
) :: {:ok, number()} | {:error, :einval}

Adjust the free parameters of the FIS (only avaliable with ANFIS engine), using the following methods:

  • Forward method: Least Square Estimate.
  • Learning method: Steepest gradient Backpropagation.
  • Energy function: 0.5 * (target - output)^2 Note: this functions fires both forward and backward passes with a batch of data.

hybrid_online_learning(pid, desired_output)

@spec hybrid_online_learning(
  atom() | pid() | {atom(), any()} | {:via, atom(), any()},
  number()
) ::
  {:ok, number()} | {:error, :einval}

Adjust the free parameters of the FIS (only avaliable with ANFIS engine), using the following methods:

  • Learning method: Steepest gradient Backpropagation.
  • Energy function: 0.5 * (target - output)^2 Note: this functions fires both forward and backward passes.

init(params)

Callback implementation for GenServer.init/1.

set_engine_type(pid, type)

@spec set_engine_type(
  atom() | pid() | {atom(), any()} | {:via, atom(), any()},
  atom()
) ::
  :ok | {:error, :einval}

Sets the Inference Engine type.

set_learning_rate(pid, learning_rate)

@spec set_learning_rate(
  atom() | pid() | {atom(), any()} | {:via, atom(), any()},
  number()
) ::
  :ok | {:error, :einval}

Sets the Learning rate (etha).

start_link(params, opt \\ [])

Spawns a Fuzzy Logic System.

The following options are require:

  • :rules - Defines the behavior of the system based on a list of rules.
  • :antecedent - (list) Defines the input variables.
  • :consequent - Defines the output variable.

stop(pid)