Syntheses of effective techniques of complex testing of technical products

Abstract

The offered method allows to define the requirements for testing measurements accuracy and optimize the cost of testing of complex electronic devices and with it to provide the required reliability of the testing results.Keywords: complex (multivariable) testing of products, simulation, testing errors of the first and second kind. IntroductionThe recommendations for a solution of the urgent problem of testing complex technical products are given in [I]. The article offers a method of optimization of complete testing of mass produced wares based on the economic criterion. 'The method implies testing several parameters which can be independent or statistically dependent (with correlation factor determined).Let us assume that the criterion of optimum product testing is the costs involved in testing C. 'These costs include the costs of testing itself and losses from testing errors. It is common practice to characterize the quality of testing (the factor of merit) by probability of a testing error of the first kind PI (risk of the manufacturer) and by probability of a testing error of the second kind P2 (risk of the consumer) [ 11.PI characterizes an average share of falsely rejected (as faulty) products in the total amount of products tested. A testing error of the first kind results from inaccuracy of testing measurements. P2 characterizes an average share of actually faulty products in the total amount of products delivered to the consumer as functioning. Errors of the second kind can result from inaccuracy of measurements and an irrational choice of number of tested parameters (see the example).The losses on account of testing errors of the second kind are losses from using malfunctioning products. Obviously, they are proportional to probability P2. These losses can be evaluated only by the consumer.We will name the testing procedure optimum if the costs of carrying out the testing are minimum and the inequality is valid:Here Pzlnax is the maximum permissible value of probability of a testing error of the second kind. The value Pzniax should be determined by both the consumer and the manufacturer of products.Further optimization is understood as choice of the best values of parameters for the testing procedure which would minimize the costs C and ensure the fulfillment of the inequality (1).The optimized parameters of the testing procedure are the permissible inaccuracy of measurements while testing and the values of checking tolerance (1). Similarly, a problem of choice of an optimum number of tested parameters can be solved. The solution of the problem.To determine the costs C let us observe the final stage of manufacturing, which includes adjustment and testing of all products. After adjustment N products comes to testing during a time interval T, of them N f are actually functioning and N ni are malfunctioning. It is supposed that malfunctioning products have latent defects, which can be revealed or not revealed while testing. 0-7803-736 1 -8/01/$17.00 02002 IEEE 118