CN115328044A - Identification model-based industrial device variable load man-machine fusion operation evaluation method - Google Patents

Abstract

The invention discloses a method for evaluating the man-machine fusion operation of industrial equipment with variable load based on an identification model. The method firstly obtains various operation data and tag data in the variable load operation of industrial devices, and then obtains the operation data of IMPC as a skill evaluation benchmark. In-process performance indicators assess operator skills; the method uses variable-load operational data from a two-tiered industrial model predictive control algorithm as a benchmark for skill assessment. The present invention can objectively reflect and distinguish the operation skill levels of different operators from multiple dimensions.

Description

Evaluation method of man-machine fusion operation with variable load of industrial equipment based on identification model

technical field

The invention relates to the technical field of process operation training of industrial devices, in particular to an identification model-based evaluation method for man-machine fusion operation of variable loads of industrial devices.

Background technique

Load adjustment is the most common operation task of process equipment, such as industrial gas production equipment, nuclear power equipment, oil refining equipment, etc. They all need to adjust the load of the equipment safely, stably and efficiently to meet the urgent needs of the downstream. For large-scale process production devices, the dynamic load adjustment process is often associated with the joint and coordinated adjustment of multiple operating variables, and the large-scale load adjustment process involves complex nonlinear operations, which requires processes without automatic load changing systems. The device operator has skilled manual load changing operation skills.

In recent years, most process plants in the chemical industry have deployed a variable load operating training system (Operator Training System, OTS). However, in the existing industrial equipment OTS system, the operation evaluation function is single, and most of them only evaluate the operator's skills from the two perspectives of the completeness of the operation steps and the correctness of the operation sequence. The function of dimension objective evaluation and guidance.

Contents of the invention

The purpose of the present invention is to address the deficiencies of the prior art, and provide an identification model-based human-machine fusion operation evaluation method for variable loads of industrial devices.

The object of the present invention is achieved through the following technical solutions: a method for evaluating the man-machine fusion operation of variable loads of industrial equipment based on identification models, comprising the following steps:

(1) Obtain various operating data and bit number data in the variable load operation of industrial devices: connect to the DCS platform of industrial devices through industrial communication protocols, obtain various production data in the production process of industrial devices in real time, and automatically record the various data of operators. item operation data;

(2) Obtain the operation data of the IMPC as the skill evaluation benchmark: design a double-layer IMPC to complete the switching task between any two working conditions in several typical working conditions, in which the double-layer IMPC consists of the upper-level steady-state optimization calculation module and the lower-level The dynamic predictive control module is composed of the dynamic predictive control module; the steady-state optimization calculation module provides the optimal tracking of the process for the lower layer, and the dynamic predictive control module controls the load adjustment process smoothly and quickly to the setting given by the upper layer without violating the constraints. Objective; After all tasks are completed, collect variable load operation data of IMPC.

(3) Evaluate the operator's skills from four performance indicators in the variable load process: safety constraints, product quality constraints, task completion time, and energy consumption;

Further, the step (3) is divided into the following sub-steps.

(3.1) Determine the DCS alarm constraints in the industrial production process.

(3.2) Determine four evaluation indicators for variable load operation skills: safety constraints, product quality constraints, task completion time, and energy consumption;

(3.3) Design safety constraint index: safety constraint is a key indicator reflecting production safety, and the score of safety constraint is recorded as T ₁ . The initial score of T1 is ₁₀₀ , and when the operator's manual operation triggers the advanced alarm constraint, the score _of T1 is gradually deducted.

(3.4) Design product quality constraint indicators: product quality constraint indicators are composed of hard product quality constraints and soft product quality constraints. Hard product quality constraints are related to the quality of industrial device products. Whether the value of key process variables triggers low-level alarm constraints To determine; soft product quality constraints are related to whether the relationship between several important materials at the end of manual operation matches each other. The steady-state material matching relationship is obtained through the following steps. First, the actual data is collected, and the steady-state working condition data is identified through the steady-state detection algorithm. Secondly, the steady-state material matching relationship is obtained by fitting the second-order polynomial model through the steady-state working condition data. Record the score of the product quality constraint as T ₂ , and the initial score of T ₂ is 100. If the operator triggers the low _- level alarm constraints of the industrial installation during the manual operation, or if the material relationship does not meet the constraint conditions at the end of the manual operation, the score of T2 will be gradually deducted.

(3.5) Design operation speed index: According to the operation data of IMPC, the task completion time index score is calculated as follows:

Among them, T _MPC and T _Operator are the time for IMPC and human operator to complete the same operation task respectively.

(3.6) Design energy consumption index: According to the operation data of IMPC, the energy consumption index score is calculated as follows:

Among them, _EMPC and _EOperator are the energy consumption of IMPC and human operator to complete the same task.

(3.7) Design a multi-index weighting method.

The final operation evaluation result T _Final is obtained by weighting the above four indicators, as follows:

Among them, q _i represents the weighting coefficient of the i-th index.

The beneficial effect of the present invention is that the method is designed to evaluate the dynamic variable load operation skills of industrial plant operators through a multi-dimensional operation skill evaluation method. The method evaluates the variable-load skills of industrial plant operators based on four performance metrics: safety constraints, product quality constraints, task completion time, and energy consumption. The method uses variable load operating data from a two-tier industrial model predictive control algorithm as the benchmark for skill assessment. It can objectively reflect and distinguish the operational skill levels of different operators from multiple dimensions.

Description of drawings

Figure 1 is the final operational skill rating radar chart.

Detailed ways

The present invention discloses an identification model-based human-machine fusion operation evaluation method for variable loads of industrial devices, which includes the following steps:

Step 1: Obtain various operating data and bit number data in the variable load operation of the industrial device: connect to the DCS platform of the industrial device through the industrial communication protocol, obtain various production data in the production process of the industrial device in real time, and automatically record the various data of the operator. item operation data;

Step 2: Obtain the operation data of IMPC as a skill evaluation benchmark: Based on the previous development of the variable load operation training system for industrial equipment, design a double-layer IMPC to complete the switching task between any two working conditions in several typical working conditions. The two-layer IMPC consists of an upper-layer steady-state optimization calculation module and a lower-layer dynamic predictive control module. The steady-state optimization calculation module provides the optimal tracking and setting goals of the process for the lower layer, and the dynamic predictive control module controls the load adjustment process smoothly and quickly to the set goals given by the upper layer without violating the constraints. After all tasks are completed, the variable load operation data of IMPC is collected, which will be used as a benchmark for manual operation skill assessment.

Step 3: Evaluate the operator's skills from the four performance indicators in the variable load process: safety constraints, product quality constraints, task completion time and energy consumption;

This step is the core of the present invention and is divided into the following sub-steps.

1) Determine the DCS alarm constraints in the industrial production process.

Combining process and historical data, determine the quality requirements in the variable load production process of industrial plants, including DCS low-level and high-level alarm constraints.

2) Determine the four evaluation indexes of variable load operation skills.

By combining process knowledge, the operator's variable load operation skills are generally evaluated from the following four perspectives: (1) operational safety; (2) product quality constraints; (3) operating speed indicators; (4) energy consumption indicators.

3) Design security constraint indicators: security constraints.

Safety constraint indicators are related to the safety of industrial installations and are determined by whether key process variable values trigger high-level alarm constraints. Advanced alarms will trigger the DCS's safety interlock system to protect industrial equipment. Safety constraints are key indicators reflecting production safety, and the score of safety constraints is recorded as T ₁ . The initial score of T1 is ₁₀₀ , and when the operator's manual operation triggers the advanced alarm constraint, the score _of T1 is gradually deducted.

4) Design product quality constraint indicators: product quality constraints.

The product quality constraint index is composed of two parts: hard product quality constraint and soft product quality constraint. Hard product quality constraints relate to the quality of industrial plant products and are determined by whether the values of key process variables trigger low-level alarm constraints. Soft product quality constraints are related to whether the relationship between several important materials matches each other at the end of manual operation. If, at the end of the manual operation, the relationship between the important materials does not match, the operating conditions of the industrial plant will not be stable in subsequent simulations.

The steady-state material matching relationship is obtained through the following steps. First, the actual data is collected, and the steady-state working condition data is identified through the steady-state detection algorithm. Note that the steady-state data here includes not only typical steady-state data, but also some atypical steady-state data. Secondly, the steady-state material matching relationship is obtained by fitting the second-order polynomial model through the steady-state working condition data.

Record the score of the product quality constraint as T ₂ , and the initial score of T ₂ is 100. If the operator triggers the low _- level alarm constraints of the industrial installation during the manual operation, or if the material relationship does not meet the constraint conditions at the end of the manual operation, the score of T2 will be gradually deducted. Different bit numbers contribute differently to hard product quality constraint indicators because of their different importance to industrial device production. Similarly, different product quality constraints are also set to contribute differently to soft product quality constraint metrics.

5) Design operation speed indicator: task completion time.

The task completion time indicator reflects the speed of response to downstream demands. While ensuring plant safety and product quality, operators need to adjust industrial unit loads as quickly as possible to meet downstream demand. According to the operation data of IMPC, the task completion time indicator score is calculated as follows:

Among them, T _MPC and T _Operator are the time for IMPC and human operator to complete the same operation task respectively.

6) Design energy consumption index: energy consumption.

The energy consumption index reflects the economy of manual operation. According to the operation data of IMPC, the energy consumption indicator score is calculated as follows:

Among them, _EMPC and _EOperator are the energy consumption of IMPC and human operator to complete the same task.

7) Design a multi-index weighting method.

This method obtains the final operation evaluation result T _Final by weighting the above four indicators, as follows:

Among them, q _i represents the weighting coefficient of the i-th indicator.

Claims (2)

1. A human-machine fusion operation evaluation method for variable loads of industrial installations based on an identification model, characterized in that it comprises the following steps: (1) Obtain various operating data and bit number data in the variable load operation of industrial devices: connect to the DCS platform of industrial devices through industrial communication protocols, obtain various production data in the production process of industrial devices in real time, and automatically record the various data of operators. item operation data; (2) Obtain the operation data of the IMPC as the skill evaluation benchmark: design a double-layer IMPC to complete the switching task between any two working conditions in several typical working conditions, in which the double-layer IMPC consists of the upper-level steady-state optimization calculation module and the lower-level The dynamic predictive control module is composed of the dynamic predictive control module; the steady-state optimization calculation module provides the optimal tracking of the process for the lower layer, and the dynamic predictive control module controls the load adjustment process smoothly and quickly to the setting given by the upper layer without violating the constraints. Objective; After all tasks are completed, collect variable load operation data of IMPC. (3) Evaluate the operator's skills from four performance indicators in the variable load process: safety constraints, product quality constraints, task completion time, and energy consumption; 2. The identification model-based human-computer fusion operation evaluation method for variable loads of industrial devices according to claim 1, wherein the step (3) is divided into the following sub-steps. (3.1) Determine the DCS alarm constraints in the industrial production process. (3.2) Determine four evaluation indicators for variable load operation skills: safety constraints, product quality constraints, task completion time, and energy consumption; (3.3) Design safety constraint index: safety constraint is a key indicator reflecting production safety, and the score of safety constraint is recorded as T ₁ . The initial score of T1 is ₁₀₀ , and when the operator's manual operation triggers the advanced alarm constraint, the score _of T1 is gradually deducted. (3.4) Design product quality constraint indicators: product quality constraint indicators are composed of hard product quality constraints and soft product quality constraints. Hard product quality constraints are related to the quality of industrial device products. Whether the value of key process variables triggers low-level alarm constraints To determine; soft product quality constraints are related to whether the relationship between several important materials at the end of manual operation matches each other. The steady-state material matching relationship is obtained through the following steps. First, the actual data is collected, and the steady-state working condition data is identified through the steady-state detection algorithm. Secondly, the steady-state material matching relationship is obtained by fitting the second-order polynomial model through the steady-state working condition data. Record the score of the product quality constraint as T ₂ , and the initial score of T ₂ is 100. If the operator triggers the low _- level alarm constraints of the industrial installation during the manual operation, or if the material relationship does not meet the constraint conditions at the end of the manual operation, the score of T2 will be gradually deducted. (3.5) Design operation speed index: According to the operation data of IMPC, the task completion time index score is calculated as follows:

Among them, T _MPC and T _Operator are the time for IMPC and human operator to complete the same operation task respectively. (3.6) Design energy consumption index: According to the operation data of IMPC, the energy consumption index score is calculated as follows:

Among them, _EMPC and _EOperator are the energy consumption of IMPC and human operator to complete the same task. (3.7) Design a multi-index weighting method. The final operation evaluation result T _Final is obtained by weighting the above four indicators, as follows:

Among them, q _i represents the weighting coefficient of the i-th index.

Images