Learn PLCs free
Intermediate20 min readMaterial Handling

Mitsubishi Structured Text for Conveyor Systems

Learn Structured Text programming for Conveyor Systems using Mitsubishi GX Works2/GX Works3. Includes code examples, best practices, and step-by-step implementation guide for Material Handling applications.

💻
Platform
GX Works2/GX Works3
📊
Complexity
Beginner to Intermediate
⏱️
Project Duration
1-3 weeks

Implementing Structured Text for Conveyor Systems using Mitsubishi GX Works2/GX Works3 requires a documented design, applicable standards, and project-specific acceptance criteria. This guide organizes practical checks for code structure, diagnostics, testing, and maintenance.

The example references FX5-family terminology, but model capabilities vary. Verify the controller manual, the installed GX Works2/GX Works3 version, and any applicable machinery, process, electrical, or functional-safety requirements for the actual project.

Best practices for Conveyor Systems encompass multiple dimensions: proper handling of 5 sensor types, safe control of 5 different actuators, managing product tracking, and ensuring compliance with relevant industry standards. The Structured Text approach, when properly implemented, provides powerful for complex logic and excellent code reusability, both critical for beginner to intermediate projects.

This guide presents a reviewable approach to Mitsubishi Structured Text programming for Conveyor Systems, covering code organization, documentation, test procedures, and maintenance handoff. The sample logic is educational and must be compiled, simulated, peer-reviewed, and tested against the project's acceptance criteria before use on equipment.

Mitsubishi GX Works2/GX Works3 for Conveyor Systems

GX Works2/GX Works3 is a programming environment associated with Mitsubishi controller families such as FX5, iQ-R, iQ-F. This guide uses Structured Text terminology from the supplied guide dataset, but controller capabilities and language support can change by model, firmware, software edition, and license.

Verify Before You Start:

  • The selected controller supports the required Structured Text constructs

  • The project version matches the installed GX Works2/GX Works3 release

  • Required communications, motion, safety, and simulation options are licensed

  • Firmware and device-description files are compatible with the project

  • The vendor manuals used for the design match the exact hardware revision


Application Planning:

For a Conveyor Systems exercise, map the required inputs and outputs before writing logic. The example considers 5 sensor types, including Photoelectric sensors, Proximity sensors, Encoders, and 5 actuator types.

Control Equipment for Conveyor Systems:

  • Belt conveyors with motor-driven pulleys

  • Roller conveyors (powered and gravity)

  • Modular plastic belt conveyors

  • Accumulation conveyors (zero-pressure, minimum-pressure)


Controller-family references used in this guide include:

  • FX5: Confirm CPU, I/O, memory, communications, and Structured Text support in the current selection guide

  • iQ-R: Confirm CPU, I/O, memory, communications, and Structured Text support in the current selection guide

  • iQ-F: Confirm CPU, I/O, memory, communications, and Structured Text support in the current selection guide

  • Q Series: Confirm CPU, I/O, memory, communications, and Structured Text support in the current selection guide


Hardware Selection Checklist:

  • Count local and remote I/O, including planned expansion

  • Measure the required task and communications update rates

  • Identify memory, data-retention, diagnostics, and cybersecurity requirements

  • Treat safety functions as a separate, standards-led design activity

  • Confirm lifecycle status, regional availability, licensing, and support


Source and Validation Note:

This page does not represent a vendor certification or a hardware acceptance test. Use current Mitsubishi manuals, release notes, and safety documentation as the authority for product-specific behavior. Validate adapted logic in a simulator or isolated test setup before connecting it to equipment.

Investment Considerations:

For Conveyor Systems projects, compare hardware, software licensing, training, engineering, test equipment, commissioning, spares, and ongoing support. Obtain current pricing and lifecycle information directly from the vendor or an authorized regional supplier.

Understanding Structured Text for Conveyor Systems

Structured Text (ST) is a high-level, text-based programming language defined in IEC 61131-3. It resembles Pascal and provides powerful constructs for complex algorithms, calculations, and data manipulation.

Execution Model:

Code executes sequentially from top to bottom within each program unit. Variables maintain state between scan cycles unless explicitly reset.

Core Advantages for Conveyor Systems:

  • Powerful for complex logic: Critical for Conveyor Systems when handling beginner to intermediate control logic

  • Excellent code reusability: Critical for Conveyor Systems when handling beginner to intermediate control logic

  • Compact code representation: Critical for Conveyor Systems when handling beginner to intermediate control logic

  • Good for algorithms and calculations: Critical for Conveyor Systems when handling beginner to intermediate control logic

  • Familiar to software developers: Critical for Conveyor Systems when handling beginner to intermediate control logic


Why Structured Text Fits Conveyor Systems:

Conveyor Systems systems in Material Handling typically involve:

  • Sensors: Photoelectric sensors for product detection and zone occupancy, Proximity sensors for metal product detection, Encoders for speed feedback and position tracking

  • Actuators: AC motors with VFDs for variable speed control, Motor starters for fixed-speed sections, Pneumatic diverters and pushers for sorting

  • Complexity: Beginner to Intermediate with challenges including Maintaining product tracking through merges and diverters


Programming Fundamentals in Structured Text:

Variables:
- declaration: VAR / VAR_INPUT / VAR_OUTPUT / VAR_IN_OUT / VAR_GLOBAL sections
- initialization: Variables can be initialized at declaration: Counter : INT := 0;
- constants: VAR CONSTANT section for read-only values

Operators:
- arithmetic: + - * / MOD (modulo)
- comparison: = <> < > <= >=
- logical: AND OR XOR NOT

ControlStructures:
- if: IF condition THEN statements; ELSIF condition THEN statements; ELSE statements; END_IF;
- case: CASE selector OF value1: statements; value2: statements; ELSE statements; END_CASE;
- for: FOR index := start TO end BY step DO statements; END_FOR;

Best Practices for Structured Text:

  • Use meaningful variable names with consistent naming conventions

  • Initialize all variables at declaration to prevent undefined behavior

  • Use enumerated types for state machines instead of magic numbers

  • Break complex expressions into intermediate variables for readability

  • Use functions for reusable calculations and function blocks for stateful operations


Common Mistakes to Avoid:

  • Using = instead of := for assignment (= is comparison)

  • Forgetting semicolons at end of statements

  • Integer division truncation - use REAL for decimal results

  • Infinite loops from incorrect WHILE/REPEAT conditions


Typical Applications:

1. PID control: Directly applicable to Conveyor Systems
2. Recipe management: Related control patterns
3. Statistical calculations: Related control patterns
4. Data logging: Related control patterns

Understanding these fundamentals prepares you to implement effective Structured Text solutions for Conveyor Systems using Mitsubishi GX Works2/GX Works3.

Implementing Conveyor Systems with Structured Text

Conveyor control systems manage the movement of materials through manufacturing and distribution facilities. PLCs coordinate multiple conveyor sections, handle product tracking, manage zones and accumulation, and interface with other automated equipment.

This walkthrough demonstrates practical implementation using Mitsubishi GX Works2/GX Works3 and Structured Text programming.

System Requirements:

A typical Conveyor Systems implementation includes:

Input Devices (Sensors):
1. Photoelectric sensors for product detection and zone occupancy: Critical for monitoring system state
2. Proximity sensors for metal product detection: Critical for monitoring system state
3. Encoders for speed feedback and position tracking: Critical for monitoring system state
4. Barcode readers and RFID scanners for product identification: Critical for monitoring system state
5. Weight scales for product verification: Critical for monitoring system state

Output Devices (Actuators):
1. AC motors with VFDs for variable speed control: Primary control output
2. Motor starters for fixed-speed sections: Supporting control function
3. Pneumatic diverters and pushers for sorting: Supporting control function
4. Servo drives for precision positioning: Supporting control function
5. Brake modules for controlled stops: Supporting control function

Control Equipment:

  • Belt conveyors with motor-driven pulleys

  • Roller conveyors (powered and gravity)

  • Modular plastic belt conveyors

  • Accumulation conveyors (zero-pressure, minimum-pressure)


Control Strategies for Conveyor Systems:

1. Primary Control: Automated material handling using conveyor belts with PLC control for sorting, routing, and tracking products.
2. Safety Interlocks: Preventing Product tracking
3. Error Recovery: Handling Speed synchronization

Implementation Steps:

Step 1: Map conveyor layout with all zones, sensors, and motor locations

In GX Works2/GX Works3, map conveyor layout with all zones, sensors, and motor locations.

Step 2: Define product types, sizes, weights, and handling requirements

In GX Works2/GX Works3, define product types, sizes, weights, and handling requirements.

Step 3: Create tracking data structure with product ID, location, and destination

In GX Works2/GX Works3, create tracking data structure with product id, location, and destination.

Step 4: Implement zone control logic with proper handshaking between zones

In GX Works2/GX Works3, implement zone control logic with proper handshaking between zones.

Step 5: Add product tracking using sensor events and encoder feedback

In GX Works2/GX Works3, add product tracking using sensor events and encoder feedback.

Step 6: Program diverter/sorter logic based on product routing data

In GX Works2/GX Works3, program diverter/sorter logic based on product routing data.


Mitsubishi Function Design:

Function block (FB) programming in Mitsubishi creates reusable logic modules with defined interfaces encapsulating complexity. FB definition includes input variables (VAR_INPUT), output variables (VAR_OUTPUT), internal variables (VAR), and retained variables (VAR_RETAIN) maintaining values between calls. Creating motor control FB: inputs include Start_Cmd (BOOL), Stop_Cmd (BOOL), Speed_SP (INT), outputs include Running_Sts (BOOL), Fault_Sts (BOOL), Actual_Speed (INT), internal variables store timers, state machine stages, and diagnostic counters. FB instantiation creates instance: Motor1 (Motor_FB) with unique variable storage, allowing multiple instances Motor1, Motor2, Motor3 controlling different motors using same logic. Array of FB instances: Motors : ARRAY[1..10] OF Motor_FB accessed as Motors[3].Running_Sts checking status of motor 3. Standard function (FUN) differs from FB by lacking internal memory, suitable for calculations or conversions: Temp_Conversion_FUN(Celsius) returns Fahrenheit without retaining historical data. Structured text programming within FBs/FUNs provides clearer logic for complex algorithms compared to ladder: IF-THEN-ELSIF-ELSE structures, FOR loops, CASE statements expressing intent more directly than ladder equivalents. EN/ENO functionality enables conditional execution: EN (enable input) controls whether FB executes, ENO (enable output) indicates successful execution detecting errors within block. Library management exports FBs to library files (.glib) shared across projects and engineering teams, versioned to track modifications and ensure consistency. The intelligent function module (IFM) templates provide pre-built FBs for common applications: PID control, analog scaling, motion positioning reducing development time and providing tested reliable code. Simulation mode tests FB logic without hardware, allowing desktop development and unit testing before commissioning. Protection functionality encrypts FB contents preventing unauthorized viewing or modification, useful for proprietary algorithms or OEM machine builders distributing programs to end users.

Common Challenges and Solutions:

1. Maintaining product tracking through merges and diverters

  • Solution: Structured Text addresses this through Powerful for complex logic.


2. Handling products of varying sizes and weights

  • Solution: Structured Text addresses this through Excellent code reusability.


3. Preventing jams at transitions and merge points

  • Solution: Structured Text addresses this through Compact code representation.


4. Coordinating speeds between connected conveyors

  • Solution: Structured Text addresses this through Good for algorithms and calculations.


Safety Considerations:

  • E-stop functionality with proper zone isolation

  • Pull-cord emergency stops along conveyor length

  • Guard interlocking at all pinch points

  • Speed monitoring to prevent runaway conditions

  • Light curtains at operator access points


Performance Metrics:

  • Task and I/O timing: Record minimum, average, and maximum values under a defined test load

  • Accuracy: Define an acceptable tolerance and compare it with calibrated reference measurements

  • Throughput: Count completed cycles over a fixed interval and record rejected or incomplete cycles

  • Fault response: Measure detection, safe-state, alarm, and recovery behavior for each test case

  • Resource use: Record memory, communications load, and diagnostic-buffer behavior

Mitsubishi Diagnostic Tools:

Device memory monitor: Real-time table displaying current values for X, Y, M, D devices with force capability,Entry data monitor: Shows actual rung logic states with contact ON/OFF indication during program execution,Device test: Manually control outputs and set internal relays for wiring verification without program influence,Intelligent module diagnostics: Buffer memory display showing module status, error codes, and configuration,Scan time monitor: Displays current, maximum, and minimum scan times identifying performance issues,Error code history: Chronological log of system errors, module faults, and CPU events with timestamps,CC-Link/network diagnostics: Visual network status showing connected stations, errors, and communication statistics,SD card operation log: Records all SD card read/write operations, file transfers, and access timestamps,Remote diagnosis via Ethernet: Connect GX Works over network for monitoring and troubleshooting without local access,Sampling trace: Records device value changes over time with trigger conditions for intermittent fault analysis,System monitor: Displays CPU load, memory usage, and battery status for predictive maintenance,Safety diagnosis (safety CPU): Dedicated diagnostics for safety I/O discrepancy detection and emergency stop chain status

Use the monitoring and diagnostic functions available in your GX Works2/GX Works3 version, and record the software, firmware, hardware, workload, and test procedure with every result.

Mitsubishi Structured Text Example for Conveyor Systems

Illustrative Structured Text example for Conveyor Systems using Mitsubishi terminology. Adapt the syntax to your GX Works2/GX Works3 release, compile it, and verify it in an isolated test environment before use on equipment.

(* Mitsubishi GX Works2/GX Works3 - Conveyor Systems Control *)
(* Structured Text Implementation for Material Handling *)
(* Mitsubishi programming supports both traditional device addressing (M0 *)

PROGRAM PRG_CONVEYOR_SYSTEMS_Control

VAR
    (* State Machine Variables *)
    eState : E_CONVEYOR_SYSTEMS_States := IDLE;
    bEnable : BOOL := FALSE;
    bFaultActive : BOOL := FALSE;

    (* Timers *)
    tonDebounce : TON;
    tonProcessTimeout : TON;
    tonFeedbackCheck : TON;

    (* Counters *)
    ctuCycleCounter : CTU;

    (* Process Variables *)
    rPhotoelectricsensors : REAL := 0.0;
    rACDCmotors : REAL := 0.0;
    rSetpoint : REAL := 100.0; (* Illustrative value; replace with a reviewed requirement *)
END_VAR

VAR CONSTANT
    (* Material Handling Process Parameters *)
    C_DEBOUNCE_TIME : TIME := T#500MS;
    C_PROCESS_TIMEOUT : TIME := T#30S; (* Illustrative value; verify for the process *)
    C_BATCH_SIZE : INT := 50; (* Illustrative value; replace with a reviewed requirement *)
END_VAR

(* Input Conditioning *)
tonDebounce(IN := bStartButton, PT := C_DEBOUNCE_TIME);
bEnable := tonDebounce.Q AND NOT bEmergencyStop AND bSafetyOK;

(* Main State Machine - Pattern: State machine implementation in Mitsubis *)
CASE eState OF
    IDLE:
        rACDCmotors := 0.0;
        ctuCycleCounter(RESET := TRUE);
        IF bEnable AND rPhotoelectricsensors > 0.0 THEN
            eState := STARTING;
        END_IF;

    STARTING:
        (* Ramp up output - Gradual start *)
        rACDCmotors := MIN(rACDCmotors + 5.0, rSetpoint);
        IF rACDCmotors >= rSetpoint THEN
            eState := RUNNING;
        END_IF;

    RUNNING:
        (* Conveyor Systems active - Conveyor control systems manage the movement of ma *)
        tonProcessTimeout(IN := TRUE, PT := C_PROCESS_TIMEOUT);
        ctuCycleCounter(CU := bCyclePulse, PV := C_BATCH_SIZE);

        IF ctuCycleCounter.Q THEN
            eState := COMPLETE;
        ELSIF tonProcessTimeout.Q THEN
            bFaultActive := TRUE;
            eState := FAULT;
        END_IF;

    COMPLETE:
        rACDCmotors := 0.0;
        (* Log production data - High-speed data logging in Mitsubishi uses file registers (R devices) organized as circular buffers with automatic SD card archiving for long-term storage. Create logging structure: file registers R0-R9999 storing 10,000 samples with each sample containing timestamp (R[base]), values (R[base+1] to R[base+10]), status (R[base+11]). Write pointer (D500) increments with each log entry: [MOV current time R[(D500*12)]] [MOV process values R[(D500*12)+1]] [INC D500] with modulo operation wrapping pointer [LD> D500 K9999] [MOV K0 D500]. Triggered logging initiates capture on alarm conditions preserving pre-trigger buffer: maintain continuous logging but flag trigger index enabling post-event retrieval of 100 samples before alarm and 500 samples after providing failure context. CSV file export uses SD card write instructions formatting file register data into comma-delimited text files readable by Excel or data analysis software: SDWR instruction writes R0-R9999 to SD:\LOG\data.csv with timestamp filename generation creating unique files daily. Sampling rates configurable from 10ms (fixed cycle interrupt program) to several minutes (main program logic) depending on process dynamics and storage capacity requirements. Data compression implements deadband filtering: log sample only when value changes exceed threshold reducing storage requirements for slowly-changing process variables like tank levels or temperatures. Integration with SCADA/historian systems uses SLMP protocol transferring logged data via Ethernet to centralized databases with automatic retry logic handling network interruptions preventing data loss. Batch correlation links production data to specific product lots: each batch start creates new log file section with batch ID header enabling traceability from raw materials through finished goods. Energy logging totalizes consumption from power meters connected via CC-Link or Modbus calculating specific energy per produced unit, identifying efficiency improvements and cost allocation by product line. Safety event logging captures all safety input states, bypass activations, and emergency stop events with tamper-proof timestamps meeting regulatory documentation requirements for incident investigations and compliance audits. *)
        eState := IDLE;

    FAULT:
        rACDCmotors := 0.0;
        (* Alarm management in Mitsubishi uses bit devices (M or B) for alarm active flags with corresponding data registers storing timestamps, values, and alarm details. Alarm structure allocates device ranges: M1000-M1999 for alarm active flags (1000 unique alarms), D5000-D5999 storing alarm timestamps or associated values. Alarm detection logic: [LD Tank_Level > High_Limit] [AND NOT previous alarm state M1000] [OUT M1000] [MOV current time D5000] capturing alarm activation moment. Alarm acknowledgment requires operator action via HMI: GOT screen button writes to acknowledgment bit (M2000) which resets alarm flag when condition clears [LD M1000] [AND alarm cleared] [AND M2000 acknowledged] [RST M1000] [RST M2000]. Priority classification uses different device ranges or separate bits: Critical alarms M1000-M1099, Warnings M1100-M1199, Information M1200-M1299 with severity-specific visual/audible HMI indicators. Alarm logging to SD card uses CSV file write instructions (SDWR) recording alarm number, timestamp, activation/deactivation, and associated process values for historical analysis and regulatory compliance. First-out alarm detection latches initial alarm in cascade of related faults: bearing temperature alarm (M1050) latches before motor overload (M1051) before production stopped (M1052) with reset sequence clearing in reverse order after root cause addressed. Integration with GOT HMI alarm viewer displays active alarms in sortable/filterable list with acknowledgment tracking, alarm help text, and corrective action guidance displayed to operators. Alarm rate limiting prevents flooding when single fault triggers hundreds of consequential alarms: introduce 5-second delays before enabling secondary alarms allowing operators to focus on root cause. Email notification for critical alarms uses Ethernet communication function blocks sending SMTP messages to distribution lists with alarm details formatted in message body. Statistical alarm analysis counts alarm frequencies storing totals in file registers: most frequent alarm identification guides preventive maintenance priorities addressing chronic equipment issues before failures occur. *)
        IF bFaultReset AND NOT bEmergencyStop THEN
            bFaultActive := FALSE;
            eState := IDLE;
        END_IF;
END_CASE;

(* Safety Override - Always executes *)
IF bEmergencyStop OR NOT bSafetyOK THEN
    rACDCmotors := 0.0;
    eState := FAULT;
    bFaultActive := TRUE;
END_IF;

END_PROGRAM

Code Explanation:

  • 1.Enumerated state machine (State machine implementation in Mitsubishi typically uses data registers (D devices) storing current state values with ladder logic evaluating state transitions. Define state constants using file registers or direct values: STATE_IDLE = 0, STATE_STARTING = 10, STATE_RUNNING = 20, STATE_STOPPING = 30, STATE_FAULT = 90 with spaced values allowing future state insertion. Current state stored in D0 with comparison instructions (LD= D0 K0) checking states and advancement logic incrementing or setting new state values (MOV K10 D0). Timer-based state transitions use timer completion contacts: [TON timer T0 reaches preset] [AND current state = STARTING] [THEN advance to RUNNING state MOV K20 D0]. Structured text programming within function blocks provides clearer state machine logic: CASE Machine_State OF 0: (IDLE state actions and transition evaluation), 10: (STARTING state actions), 20: (RUNNING state actions), END_CASE. State transition logging writes state changes to file registers creating audit trail: when D0 changes, copy timestamp and previous state to R[pointer] incrementing pointer for circular buffer. Fault handling sets state to FAULT (90) with fault code stored in separate register (D10) indicating cause: 100=Emergency_Stop, 101=Overload, 102=Communication_Loss, with reset logic evaluating fault code and returning to appropriate safe state. Parallel state machines coordinate through shared status bits: Machine1_State (D100), Machine2_State (D200) with interlock logic preventing conflicting operations. HMI integration displays state names using text switching on D0 value converting numeric states to operator-friendly descriptions: 0='Ready', 10='Starting', 20='Running'. Function block encapsulation creates reusable state machine logic instantiated multiple times for identical equipment with instance-specific state storage in FB internal variables.) for clear Conveyor Systems sequence control
  • 2.Constants use clearly marked illustrative values that must be replaced with reviewed project requirements
  • 3.Input conditioning with debounce timer prevents false triggers in industrial environment
  • 4.STARTING state implements soft-start ramp - prevents mechanical shock
  • 5.Process timeout detection flags a possible stuck condition for investigation
  • 6.The final override illustrates a software permissive only; it is not a safety-rated function and must not replace a validated safety system

Best Practices

  • Follow Mitsubishi naming conventions: Mitsubishi programming supports both traditional device addressing (M0, D100, X1
  • Mitsubishi function design: Function block (FB) programming in Mitsubishi creates reusable logic modules wit
  • Data organization: Mitsubishi uses file registers (R devices) and structured data in function block
  • Structured Text: Use meaningful variable names with consistent naming conventions
  • Structured Text: Initialize all variables at declaration to prevent undefined behavior
  • Structured Text: Use enumerated types for state machines instead of magic numbers
  • Conveyor Systems: Use rising edge detection for sensor events, not level
  • Conveyor Systems: Implement proper debouncing for mechanical sensors
  • Conveyor Systems: Add gap checking before merges to prevent collisions
  • Debug with GX Works2/GX Works3: Use sampling trace to capture high-speed events occurring faster than
  • Safety: E-stop functionality with proper zone isolation
  • Use a compatible simulator or isolated test rig to test Conveyor Systems logic before deployment

Common Pitfalls to Avoid

  • Structured Text: Using = instead of := for assignment (= is comparison)
  • Structured Text: Forgetting semicolons at end of statements
  • Structured Text: Integer division truncation - use REAL for decimal results
  • Mitsubishi common error: Error 2110: Illegal device specified - accessing device outside configured range
  • Conveyor Systems: Maintaining product tracking through merges and diverters
  • Conveyor Systems: Handling products of varying sizes and weights
  • Neglecting to validate Photoelectric sensors for product detection and zone occupancy leads to control errors
  • Insufficient comments make Structured Text programs unmaintainable over time

Related Certifications

🏆Mitsubishi PLC Programming Certification
🏆Advanced Mitsubishi Programming Certification

Applying Structured Text to Conveyor Systems using Mitsubishi GX Works2/GX Works3 requires understanding the platform, the process, and the project's acceptance criteria. This guide has covered implementation structure, an illustrative code example, verification practices, and common pitfalls for a beginner to intermediate Conveyor Systems exercise.

Use the practices outlined here to create a design that can be reviewed and tested. Define performance targets in the project requirements and confirm them with repeatable measurements.

Next Steps:

1. Check Sources: Read the current GX Works2/GX Works3 help, controller manual, release notes, and relevant standards
2. Practice Safely: Adapt the example in a simulator or isolated training setup
3. Review: Have the I/O map, state behavior, faults, and recovery steps reviewed
4. Test: Record normal, boundary, fault, restart, and communications test results

Structured Text Foundation:

Structured Text (ST) is a high-level, text-based programming language defined in IEC 61131-3. It resembles Pascal and provides powerful constructs for...

Project duration depends on scope, reviews, hardware availability, software and firmware versions, testing, commissioning, and site constraints. Remember: Use rising edge detection for sensor events, not level

For further learning, explore related topics including Recipe management, Warehouse distribution, and Mitsubishi platform-specific features for Conveyor Systems optimization.